[12:04] <lamont__> doko: I'll take a look and see what I can do
[06:31] <fabbione> morning
[07:21] <desrt> fabbione; do you still need access to gorecki?
[07:21] <fabbione> desrt: nope.. thanks :)
[07:22] <desrt> awesome
[07:22] <fabbione> desrt: i am confident you did try the new headers.. didn't you?
[07:23] <desrt> fabbione; actually, no
[07:23] <desrt> fabbione; i've been avoiding too much kernel hacking so that i didn't accidentally crash the box when you needed to use it :)
[07:23] <fabbione> oh 
[07:23] <fabbione> thanks :)
[07:23] <fabbione> well crash it and let me know :)
[07:23] <desrt> k.  lemme check
[07:24] <fabbione> i need to go offline soon for approx 20 minutes
[07:24] <desrt>   MODPOST
[07:24] <desrt> modpost: /home/desrt/coconet/coconet.o no symtab?
[07:25] <desrt>  /bin/sh: line 1: 14814 Aborted                 scripts/mod/modpost -m -a -i /usr/src/linux-headers-2.6.12-3-powerpc64-smp/Module.symvers /home/desrt/coconet/coconet.o /home/desrt/coconet/large-page.o
[07:25] <desrt> very possibly my fault :)
[07:25] <fabbione> yes
[07:25] <fabbione> i saw that too
[07:25] <fabbione> but given that all the rest compiles...
[07:26] <fabbione> i sort of assumed it was a specific coconet problem
[07:26] <desrt> let me check
[07:26] <desrt> no.  it's a -headers problem, i think
[07:27] <desrt> i'll try and find out more about it while you're offline
[07:27] <fabbione> i don't think it is
[07:27] <desrt> (compiles just fine against the full kernel source 2.6.12 with ubuntu patches)
[07:27] <fabbione> hmmmm
[07:27] <fabbione> ok... leave my account around..
[07:27] <fabbione> i will look at it later
[07:28] <fabbione> desrt: did you actually upgrade the kernel and headers to the latest version or are you just using the one i hacked locally?
[07:29] <desrt> whatever is currently in /usr/src/linux-kernel-headers
[07:29] <fabbione> currently as in?
[07:29] <fabbione> currently can be that you dist-upgraded the box 2 minutes ago
[07:29] <fabbione> or currently is what i left there a few days back :P
[07:29] <desrt> i just did dist-upgrade now.. the only difference was a new yaboot
[07:29] <fabbione> yeah the yaboot we fixed together 
[07:30] <desrt> nod.  i'm only running ubuntu-base... so not many updates
[07:30] <fabbione> ok gimme a bit and i will come back
[07:43] <desrt> ok.. i've found the problem.. and it's one of those big sort of problems that won't easily go away
[07:43] <fabbione> desrt: re
[07:43] <desrt> /usr/src/linux-headers-2.6.12-3-powerpc64-smp/scripts is a symlink to linux-headers-2.6.12-3/scripts
[07:44] <desrt> ie: all the various different arch versions share the same scripts/ directory
[07:44] <fabbione> that's right...
[07:44] <desrt> the modpost script has a compile-time flag for 32bit or 64bit mode
[07:44] <desrt> if you run the 32bit version on a 64bit .o file you get the "missing symtab" error
[07:44] <desrt> it gets its settings from elfconfig.h
[07:45] <desrt> which contains this: #define KERNEL_ELFCLASS ELFCLASS32
[07:45] <desrt> you'll basically need 2 versions of the modpost script... one for 32bit kernels and one for 64bit
[07:46] <fabbione> crack
[07:47] <fabbione> we need a completely different approach here :/
[07:47] <fabbione> bah
[07:47] <fabbione> ok.. i got the problem thanks...
[07:47] <desrt> thanks again for all your work on this :)
[07:48] <fabbione> this is going to be a real problem
[07:48] <desrt> btw: there is a working version of the kernel source in /usr/src/linux-source-2.6.12 that you can use for comparison
[07:48] <fabbione> nah i got it thanks
[07:48] <desrt> if you compile coconet the top line in the Makefile is LINUX=/path/to/source
[07:48] <fabbione> this is going to cost me an extra amount of painful work
[07:49] <desrt> anything i can do to help?
[07:49] <fabbione> desrt: not really.. this needs to be fixed in the source package...
[07:49] <desrt> (just think -- you're solving the problem for biarch amd64 too) :)
[07:49] <fabbione> when we build the kernel on the buildd's
[07:49] <fabbione> that's the real problem
[07:49] <fabbione> desrt: nope.. ppc64 is a "special case" all over the Makefiles
[07:50] <desrt> ah
[07:50] <fabbione> because it's the only one that's cross compiled
[07:50] <daniels> god I hate building the monolith
[07:50] <daniels> even with it not building anything in programs/ except for the server and most of lib/
[07:50] <daniels> hate hate hate
[07:50] <fabbione> don't worry.. the machine i use to build is fast :)
[07:50] <desrt> daniels; Xc?
[07:50] <daniels> desrt: yeah
[07:51] <desrt> i really like the fd.o broken-out-everything approach with pkgconfig help
[07:51] <fabbione> my back is hurting a lot
[07:51] <fabbione> crap i must have slept really bad this night
[07:52] <desrt> clearly an indication that you require a mid-morning nap :)
[07:52] <fabbione> desrt: i am not just tired.. my back is really hurting a lot
[07:54] <daniels> desrt: you're telling me
[04:14] <mirak> hello
[04:14] <mirak> I am trying to cross compile
[04:14] <mirak> my two architectures are ppc and i386
[04:14] <mirak> I tried the howto
[04:15] <mirak> but it both fail when I try to build gcci386 for ppc and gcc-ppc for i386
[04:16] <jbailey> Which howto?
[04:16] <jbailey> If it was anything other than http://kegel.com/crosstool/ you'd probably do better to start for there.
[04:49] <mirak> jbailey /usr/share/doc/toolchain/howto
[04:49] <mirak> I use this how to
[04:50] <mirak> I already followed the http://kegel.com/crosstool/ 
[04:50] <mirak> some month ago but it failed as well
[04:50] <mirak> I use tpkg-make
[07:04] <karim> hi
[07:04] <karim> how important are the glibc version and kernel version for cross compiling
[07:04] <karim> ?
[07:05] <karim> jbailey: hey
[07:05] <jbailey> The versions shouldn't really matter.  Cross compiling on older stuff might be a bit more finicky but not much.
[07:05] <jbailey> You generally want to make sure your libraries that you're using for cross compiling are the same or older than the ones you'll be executing with.
[07:05] <jbailey> Otherwise you risk symbol mismatches and stuff.
[07:08] <karim> here is what I want to do
[07:08] <karim> I want to set up a cross compile gcc 
[07:08] <karim> to use with distcc
[07:08] <karim> my goal is to use this ubuntu i386 box to help this poor ppc G3
[07:09] <karim> so I don't think I need all the gdb and stuff to do that
[07:09] <karim> jbailey: I don't understand why toolchain tpkg-make script fails 
[07:09] <karim> it's a pity because it's very handy
[07:10] <jbailey> I haven't used that script, so I don't know.
[07:10] <jbailey> I have some vague dreams of making it so that building cross compilers from the standard toolchain packages is trivial, but I haven't taken them past the dream stage.
[07:17] <karim> well it IS trivial
[07:17] <karim> but the compile fail
[07:17] <karim> a problem with some files missing it seems
[07:17] <karim> signal.h
[07:31] <jbailey> IIRC, signal.h means you don't have glibc headers in place.
[07:41] <zul|working> someone should start an arm port..
[07:45] <jbailey> Planning on it.
[07:57] <zul|working> wasnt there a paper at debconf for porting debian to another arch?
[07:57] <jbailey> Dunno.  I've done it twice, so wouldn't need much of a paper.
[08:12] <karim> jbailey: how do I put them in place ?
[09:06] <jbailey> karim: It depends where you've configured gcc to look for them.
[09:06] <jbailey> I tend to use --with-sysroot
[09:09] <karim> hem I am trying to use distcc
[09:09] <karim> I can remotely connect and compile
[09:09] <karim> but I don't know how to tell distccd to use the ppc gcc cross compiler
[09:13] <jbailey> Dunno.  I've read about doing that but have never done so.
[09:13] <jbailey> I thought about doing that for the Debian hurd-i386 buildd. =)
[09:14] <jbailey> Have a huge ccache tied in with distcc and cross compilers.
[09:14] <jbailey> Make the hurd the fastest compile farm in Debian. =)
[09:14] <jbailey> Of course, doko might kill me for thinking that thought out loud. =)
[09:14] <doko> ?
[09:15] <doko> so you did fix the biarch i386 builds?
[09:15] <doko> ;)
[09:15] <jbailey> karim: Or he'll play dirty. =)
[09:16] <karim> what ?
[09:25] <karim> I give up
[09:42] <karim> if I compile for ppc on the i386 I can execute the binary on the ppc host
[09:42] <karim> so cross compile work
[09:42] <karim> but however distcc fail
[09:42] <karim> it produce a little binary
[09:43] <karim> that can't be executed
[10:02] <doko> lamont, lamont-away: 4.0 did fail on the ia64 and hppa unstable buildds
[10:02] <lamont> figures
[10:19] <jbailey> Eh?  What's different than in breezy?
[10:23] <doko> nothing is different. the ia64 buildd looks seriously broken