[01:03] <airlied> BenC: pong
[04:37] <bluefoxicy> when you all get back from #-meeting in an hour, does anyone want to help me figure out wtf as I try to patch something into my kernel?
[04:37] <bluefoxicy> arch/i386/kernel/process.c:667: error: per_cpu__cpu_gdt_table undeclared (first use in this function)
[04:37] <bluefoxicy>                 load_user_cs_desc(cpu, next_p->mm); // the line in question, this is a macro
[04:38] <bluefoxicy> #define load_user_cs_desc(cpu, mm) per_cpu(cpu_gdt_table, (cpu))[GDT_ENTRY_DEFAULT_USER_CS]  = (mm)->context.user_cs  // The macro's definition
[05:54] <BenC> wow, I had no idea how much space the debian/abi directory was taking up
[05:55] <BenC> # du -sh debian/abi
[05:55] <BenC> 119M    debian/abi
[05:55] <zul> oi
[05:57] <BenC> and there it's done
[05:58] <zul> yay...
[05:58] <zul> now you can go play poker in theory
[06:00] <bluefoxicy> BenC:  what does abi do?
[06:01] <BenC> abi is the kernel compatibility
[06:01] <maks> abi does not exist ;)
[06:01] <BenC> if the ABI changes, then modules compiled against it (like LRM) have to be recompiled
[06:01] <BenC> if the ABI doesn't change, then all modules should still work
[06:02] <BenC> maks: no, there's no _stable_ abi, but there is an abi :)
[06:02] <maks> BenC: catched ;)
[06:06] <fabbione> BenC: congrats boy
[06:07] <fabbione> BenC: time to fork ubuntu-dapper-2.6 and start ubuntu-2.6 for edgy? ;)
[06:07] <BenC> fabbione: don't forget everyone that helped, like you :)
[06:07] <fabbione> i didn't do much come on
[06:07] <BenC> yep, that's my unofficial Friday excercise
[06:07] <fabbione> ehehe
[06:07] <fabbione> BenC: once you open edgy can you please merge gfs2 from kernel.org?
[06:08] <BenC> gfs2 git tree?
[06:08] <fabbione> i want an early start for the cluster stuff since it will change a lot
[06:08] <fabbione> yes
[06:09] <fabbione> we will be able to kill cman from the kernel
[06:09] <fabbione> it's all userland now
[06:09] <fabbione> otherwise i can prepare a commit that will revert all the cluster bits from GFS1
[06:10] <fabbione> and we can merge gfs2 later
[06:14] <zul> BenC, then you can show me some stuff later
[06:26] <BenC> zul: sure thing
[08:28] <BenC> setuid: if you just want to build stock kernels for Ubuntu, without initrd, just do a normal kernel build, install, modules-install
[08:28] <BenC> and edit /boot/grub/menu.lst to reflect the lack of initrd
[08:29] <setuid> Of course, but I need the new kernel to represent the same level of functionality as the running Ubuntu kernels, so if that's as simple as dropping in a .config from /boot, I'm golden. I remember running into all kinds of weirdness with this approach about 2 months ago. 
[08:29] <BenC> Probably just adding a custom entry to grub's menu for your kernel would suffice (and keep you from messing up anything specific to Ubuntu's stuff)
[08:29] <setuid> I'll give it a go again, maybe bugs were fixed. 
[08:29] <setuid> Naturally
[08:29] <BenC> copy the config
[08:29] <BenC> but without an initrd, you will need to change some drivers to be built-in (storage drivers, and filesystem drivers)
[08:31] <BenC> if you want to build the proper way "make-kpkg --initrd --rootcmd fakeroot kernel_image"
[08:33] <BenC> setuid: I have one question, if you don't like the way that Ubuntu's kernel images are working (the boot process, initrd, etc.), why are you using Ubuntu for testing of these drivers?
[08:38] <setuid> BenC: Trying to model the environment with newer kernels for our users (I'm the maintainer of pilot-link, which projects like Evolution, gnome-pilot, J-Pilot, KPilot, PilotManager and several commercial projects rely upon) 
[08:38] <setuid> So we get reports of bugs, and I need to identify if they're due to kernel patches, or kernel bugs. 
[08:38] <setuid> Then flet out from there
[08:38] <setuid> Fedora is the worst offender, I'm afraid to admit. 
[08:39] <setuid> 9 times out of 10, going to a non-Fedora kernel solves a truckload of users problems with sync 
[08:43] <setuid> make silentoldconfig doesn't seem to work with the most-recent 6 2.6.16 kernels
[08:43] <setuid> I'll keep going back 
[08:44] <desrt> silentoldconfig is only a good idea if you're actually using the same kernel version as the config was generated for
[08:44] <desrt> since otherwise there will be deltas in the available options and you probably want to use 'oldconfig'
[08:44] <setuid> Or close enough, but yes. 
[08:44] <setuid> I'll mrproper and restart
[08:44] <setuid> I wish mconfig supported 2.6
[08:48] <desrt> does 2.7 exist yet?
[08:50] <setuid> Some work in git
[08:50] <setuid> iirc
[09:15] <crimsun> BenC: thanks again!
[09:32] <chuck> heylo
[09:32] <chuck> doh
[09:37] <crimsun> hey zul :-)
[09:37] <zul> hey crimsun 
[09:41] <zul> so I take it im going to the xen conference thingy
[09:42] <crimsun> awesome
[09:43] <zul> heh i was asking
[10:29] <setuid> Ok, that didn't work (blew up in initrd, as I suspected)
[10:29] <setuid> Why does mkinitrd make a cramfs initrd, instead of a standard compressed cpio image? 
[10:30] <setuid> I've taken a vanilla 2.6.16.16 kernel and copied my Ubuntu's 2.6.15-22 .config into the tree, did make-kpkg kernel-image, it built the kernel + modules... but initrd fails. 
[10:30] <mjg59_> Because it's mkinitrd, not mkinitramfs
[10:30] <setuid> So I need to use mkinitramfs? 
[10:31] <setuid> But the resulting line in grub's config is for an initrd? How... confusing. ;) 
[10:31] <mjg59_> The filename is still initrd for compatibility reasons
[10:31] <mjg59_> It's loaded in the same way
[11:12] <setuid> Almost worked, it booted and X failed, probably some fglrx thing.. no r300 or radeon in this tree, apparently.