[09:38] <AnAnt> I am recompiling the kernel is that I am applying a patch for MMC v4 device support , if that works, shall I upload the new source/patch somewhere in MOTU or so ?
[09:41] <ivoks> no, MOTU doesn't have anything to do with kernel
[09:41] <ivoks> you should make your patch available somewhere and make a notice on kernel-devel or here...
[09:44] <AnAnt> btw, it's not my patch, it already exists in 2.6.16
[09:45] <ajmitch> then it'll be in 2.6.17, which will be in edgy
[09:46] <AnAnt> ok
[09:46] <AnAnt> you don't add features to dapper, right ? just security fixes
[09:46] <ajmitch> correct
[09:47] <ajmitch> security & critical bug fixes
[09:47] <AnAnt> can I get the 2.6.17 from dapper-backports when u guys start working on edgy ?
[09:47] <ajmitch> no, backporting the kernel would be quite a maintenance hassle
[09:48] <ajmitch> as far as I know, there are no plans to do that
[09:48] <AnAnt> k
[09:51] <kimo> Hey, ubuntu official kernel has ipw2200 ver 1.1, which is newer than the one in 2.6.16 ?? (does this make sense) ?
[09:52] <AnAnt> oh btw, why are there additions in kernel 2.6.15 (by Ubuntu) that aren't in 2.6.16 ? don't the kernel guys trust it ?
[09:54] <kimo> :)
[09:54] <kimo> any idea when are we getting a kernel update for dapper 
[09:55] <kimo> it still doesnt poweroff my laptop :)
[11:27] <AnAnt> help I get this error when compiling the kernel source: "Error in source file, line 35" , I get this from drivers/usb/net/zd1211/zddevlist.h
[03:59] <tuxmaniac> Has there been any success stories with BroadCom 4318 802.11 Card?
[05:09] <alex_joni> hello, I get an error during make-kpkg on a linux-source-2.6.15 (compiler & helper programs from apt-get build-dep linux-source-2.6.15) : Error(/usr/src/linux-source-2.6.15//kernel/printk.c:516): cannot understand prototype: 'ipipe_spinlock_t __ipipe_printk_lock = IPIPE_SPIN_LOCK_UNLOCKED; '
[05:43] <alex_joni> ok, found the problem (RTAI patch altered the comment inside kernel/printk.c)
[05:46] <crimsun> ah, I didn't read that you patched it
[05:47] <crimsun> (this connection is completely buggered)
[05:47] <alex_joni> crimsun: sorry, might have not been very clear on that :)
[05:48] <alex_joni> I see the dafault compiler is gcc 4.0, I used to use 3.x for kernel builds.. is the 4.0 ok?
[05:49] <crimsun> I can't speak for RTAI, but 4.0 is fine in dapper, yes
[05:49] <alex_joni> ok, thanks
[05:49] <crimsun> or whichever parts of 4.0.3 are used
[05:51] <mkrufky> BenC: yt?  I am reading bug # 33096 ... This is a result of your commit: 30fbd96c515f7ad612486dbd94099b0a461e406a ... Your commit message references an UbuntuBug 5773, but I am unable to locate that...  Anyhow, It looks to me as if your changeset is wrong.... If you can somehow confirm bug #5773 (missing on launchpad.net) , then I would be happy to fix this in upstream
[05:52] <BenC> 5773 is probably a bugzilla bug
[05:52] <mkrufky> ah, i'll take a look
[05:52] <mkrufky> not on bugzilla.kernel.org
[05:53] <mjg59> bugzilla.ubuntu.com
[05:53] <BenC> bugzilla.ubuntu.com
[05:53] <mkrufky> lol , silly me
[05:54] <alex_joni> any of you uses ccache when compiling new kernels?
[05:56] <mkrufky> ok, i see 5773, and this looks reasonable to me
[05:57] <mkrufky> one strange thing is that a message dated "2005-08-23" says "This bug has been fixed a long while ago, but for some reasons we forgot to close it." .... but the -git commit is dated "12 Dec 2005"
[05:57] <mkrufky> anyhow, I'll fix this in upstream for 2.6.18 , and i'll note it on the launchpad.net bug
[06:01] <zul> meh...stupid typos
[06:07] <alex_joni> any idea what : drivers/built-in.o: In function `imacfb_probe':imacfb.c:(.init.text+0xc91): undefined reference to `efi' means?
[06:07] <mjg59> It means you're trying to build a kernel without efi support?
[06:07] <alex_joni> mjg59: there should be a CONFIG_EFI or similar?
[06:08] <mjg59> Yes
[06:08] <zul> mmmm...champisonhip manager
[06:08] <alex_joni> mjg59: I started with a .config from a 2.6.12, and used make oldconfig.. is that bad?
[06:09] <mjg59> Yes. Just enable EFI.
[06:09] <alex_joni> ok, thanks
[08:34] <alex_joni> anyone knows where the bootsplash is in ubuntu?
[08:34] <alex_joni> I built a new kernel, which boots alright, but it stay black during the bootphase
[08:36] <crimsun> that's usplash, which is userspace
[08:36] <crimsun> bootsplash isn't in the kernel at all
[08:37] <alex_joni> err, sorry.. that's what I meant.. is there a special .config entry I need to take care of (FB or defaultresolution or something like that?)
[08:38] <ajmitch> morning crimsun 
[08:38] <crimsun> 'morning ajmitch 
[08:38] <alex_joni> crimsun: the stock 2.6.15 works as it should, just my fresh built kernel doesn't
[08:38] <crimsun> alex_joni: there are a number of CONFIG_FB{,_*} enabled as 'y', yes
[08:40] <infinity> We build all the FB stuff modular, actually, but having it builtin would work just as well.
[08:40] <alex_joni> crimsun: checked them (they are similar to the ones in the stock config )
[08:40] <alex_joni> most of them 'm' some 'y'
[08:41] <infinity> alex_joni: You need (at this point) either vga16fb or vesafb to make usplash happy.
[08:41] <alex_joni> I just diffed my config against the stock one, and I only added one line (related to RTAI), the rest is the same
[08:41] <infinity> vesafb will only work if you're booting with vga=XXX on the command line, otherwise you want vga16fb (which is what we default to)
[08:41] <crimsun> alex_joni: did you regenerate the initramfs?
[08:42] <alex_joni> crimsun: I used make-kpkg to make a new kernel deb and installed that. is that enough?
[08:42] <infinity> CONFIG_FB_VGA16=m
[08:42] <infinity> CONFIG_FB_VESA=m
[08:42] <ajmitch> heh, morning infinity 
[08:42] <alex_joni> infinity: both set
[08:42] <crimsun> alex_joni: infinity knows this area better than I do. Did you use --initrd ?
[08:43] <alex_joni> yes
[08:43] <alex_joni> make-kpkg --initrd --revision=aj1 --stem=linux binary-arch
[08:43] <crimsun> I believe you need to exec update-initramfs, then
[08:43] <infinity> crimsun: make-kpkg SHOULD be popping the update-initramfs call in the kernel postinst.
[08:44] <infinity> Unless we messed that up for custom kernels, and we're only doing it on stock builds.... Which would suck.
[08:44] <infinity> And would warrant an upload to -updates, IMO.
[08:44] <alex_joni> infinity: any way to check?
[08:44] <crimsun> grep initramfs /var/lib/dpkg/info/linux-image-$(uname -r).postinst
[08:44] <infinity> grep update-initramfs /var/lib/dpkg/info/linux-image-2.6.15-23-686.postinst
[08:45] <infinity> (replacing with your package name)
[08:45] <infinity> Which, if you didn't use --stem=linux, is probably "kernel-image-$foo"
[08:45] <alex_joni> I did use --stem=linux
[08:45] <infinity> Kay, then linux-image-$foo. :)
[08:46] <alex_joni> it says: "my $update_initramfs = "/usr/sbin/update-initramfs";"
[08:46] <alex_joni> so I guess it does get called
[08:47] <infinity> Should do, yes.
[08:47] <infinity> You're booting with "splash" on the command line>
[08:47] <infinity> s/>/?/
[08:47] <alex_joni> yes
[08:48] <alex_joni> same grub/menu.lst entry as for the stock 2.6.15-23-386
[08:49] <infinity> Can you toss me your .config?
[08:49] <alex_joni> infinity: sure
[08:49] <infinity> This may be easier to debug if I just see it in action.
[08:49] <infinity> Or, alternately, toss me your kernel packages.
[08:49] <alex_joni> http://www.robcon.ro/config-2.6.15-magma
[08:49] <infinity> I trust that you won't trojan then in the next 5 seconds.
[08:49] <infinity> s/then/them/
[08:50] <alex_joni> it might take me more than 5 seconds to upload them ;) hang on
[08:51] <infinity> Yeah, I see no obvious reason why this config should hate you.
[08:51] <infinity> update-initramfs -u -k 2.6.15-magma
[08:51] <infinity> That should regenerate the initramfs (again)
[08:51] <infinity> But I'm curious about why it's breaking in the first place, so a copy of your package might be nice.
[08:52] <alex_joni> they are on their way.. 
[08:52] <alex_joni> http://dsplabs.cs.upt.ro/~juve/tempdebs/ - but it will be a short while till they make it all there
[08:53] <alex_joni> infinity: need the source too? 
[08:54] <infinity> Nah.
[08:55] <alex_joni> doc is finished,  headers & image next
[08:56] <zul> brb...new kernel
[09:17] <alex_joni> infinity: done, they are there
[09:18] <infinity> .ro ... You're not some l33t Romanian script kiddie trying to eat my computer, right? :)
[09:18] <alex_joni> infinity: I surely hope not ;)
[09:18] <infinity> (Why are there so many script kiddies in .ro?  I've not yet figured this out)
[09:18] <alex_joni> neither have I.. too much spare time?
[09:19] <infinity> Could be.  Someone needs to get them all involved in constructive hacking.
[09:19] <infinity> Or, show them the outside world.  Wichever.
[09:19] <alex_joni> I tried.. but failed :)
[09:19] <infinity> "Look, there's sun and grass and stuff out there!"
[09:20] <alex_joni> I think option #2 is more appropriate.. (except that right now it's raining cats & dogs)
[09:20] <infinity> I rather like cats, but I suppose at high velocity they may be painful.
[09:21] <alex_joni> it does help that they land on their feet each time..
[09:21] <infinity> Not when that means claws on your head.
[09:22] <alex_joni> actually I think they extend their claws only afterwards.. on first impact they should be retracted :D
[09:23] <infinity> Okay, your kernel image is installed.
[09:23] <infinity> Time to reboot and get my machine hX0red. :)
[09:24] <alex_joni> infinity: btw, thanks for looking at this..
[09:38] <alex_joni> mjg59: still around? I seem to have run again into a problem with efi..
[09:41] <infinity> alex_joni: Okay, that was informative.  Spotted the difference.
[09:41] <alex_joni> infinity: what is it?
[09:41] <infinity> From our package build scripts:
[09:41] <infinity> if [ -f kernel/drivers/video/vesafb.ko ] ; then
[09:41] <infinity>         ln kernel/drivers/video/vesafb.ko initrd
[09:41] <infinity> fi
[09:42] <infinity> make-kpkg doesn't do that.
[09:42] <alex_joni> oh, I see
[09:42] <infinity> And initramfs-tools is specificall sanning the initrd directory to see if you have framebuffers there you want to include.
[09:42] <infinity> Err, with the mkdir even:
[09:42] <infinity> mkdir initrd
[09:42] <infinity> if [ -f kernel/drivers/video/vesafb.ko ] ; then
[09:42] <infinity>         ln kernel/drivers/video/vesafb.ko initrd
[09:42] <infinity> fi
[09:42] <alex_joni> ok, so what do you advice?
[09:43] <infinity> There.  So, you can just do that by hand in /lib/modules/$(uname -r) and regenerate your initramfs, and you'll be golden.
[09:43] <infinity> Or, stop using vga=XXX on your command line, cause vga16fb will work fine with your kernel.
[09:43] <alex_joni> infinity: I need to produce debs that will be distributed
[09:43] <infinity> We should probably fix make-kpkg to do the initrd/ thing for custom kernels, but that's not going to happen in dapper.
[09:44] <alex_joni> hmm, afaik this worked in breezy
[09:44] <infinity> Did you ever actually test with "vga=XXX" in  breezy?
[09:44] <infinity> Without it, this would be working fine.
[09:44] <alex_joni> but I don't have vga=XXX.. do I?
[09:44] <infinity> You almost certainly do.
[09:45] <infinity> Otherwise, you'd not be seeing the black screen.
[09:45] <alex_joni> any idea where the vga=XXX could come from?
[09:45] <infinity> cat /proc/cmdline
[09:45] <infinity> Mine, for example:
[09:46] <infinity> root=/dev/sda3 ro quiet splash vga=0x343
[09:46] <infinity> That last bit says "use vesafb, or give me a useless black screen if it's not available!"
[09:46] <alex_joni> root=/dev/hdb9 ro quiet splash
[09:46] <infinity> And it's not available, cause initramfs-tools isn't copying it into the initramfs.
[09:46] <infinity> Okay, that's bizarre...
[09:46] <alex_joni> no vga here.. :-/
[09:47] <infinity> Let me reboot again, but I'm pretty sure your kernel should worh with vga16fb...
[09:47] <alex_joni> another topic, when I remove ACPI support (because RTAI conflicts with it), I get build errors... any ideea about something like this?
[09:56] <infinity> Oh, I am so awesome, it hurts. ?/
[09:56] <alex_joni> infinity: don't tell me you nailed it?
[09:57] <infinity> alex_joni: Because initramfs-tools isn't finding any framebuffers in /initrd/, it's not loading fbcon.
[09:57] <infinity> No fbcon, no usplash.
[09:57] <infinity> And yes, this behaviour has changed since breezy, since I'm an idiot.
[09:57] <infinity> Sort of.
[09:57] <alex_joni> ok.. so basicly a make-kpkg bug?
[09:57] <infinity> Well, a me bug.
[09:58] <alex_joni> infinity: happens to all of us ;)
[09:59] <alex_joni> usually on this: http://www.linuxcnc.org/
[10:00] <infinity> Now, the trick to fix this in your custom images is to do the following:
[10:00] <infinity> Take the snippet I pasted above, and stick it in a shell script called "post-install"
[10:01] <infinity> And then cp post-install your-linux-source-tree/debian/post-install before you do the make-kpkg thing.
[10:01] <alex_joni> oh, ok
[10:01] <alex_joni> trying now
[10:01] <infinity> Give or take.
[10:01] <infinity> This is untested. :)
[10:02] <infinity> You probably want this before it:
[10:02] <infinity> cd "$IMAGE_TOP/lib/modules/$version"
[10:03] <alex_joni> do I make it a proper shell script? #!/bin/sh on top?
[10:03] <infinity> If "dpkg-deb -c blah.deb" shows a /lib/modules/`$uname -r`/initrd directory with vesafb.ko in it when you're done, you win.
[10:04] <infinity> alex_joni: Yes, a proper shell script.
[10:04] <infinity> Exectuable, too. :)
[10:04] <alex_joni> done
[10:04] <alex_joni> now if I could only make it build :(
[10:04] <infinity> It gets called by make-kpkg after installing all the junk in the image directory, but before generating the .deb
[10:04] <infinity> Our Ubuntu post-install is a bit more involved, but that's one of the things it does.
[10:04] <alex_joni> ok, thanks for the fix 
[10:05] <alex_joni> any idea about how to make it compile with ACPI unset?
[10:06] <infinity> That, I have no idea about.
[10:06] <infinity> Or, rather, I can't beging to have a clue given your "I get build errors" synopsis above.
[10:06] <alex_joni> I can post a better error message
[10:07] <alex_joni> arch/i386/kernel/built-in.o: In function `setup_arch': undefined reference to `check_acpi_pci'
[10:07] <alex_joni> drivers/built-in.o: In function `imacfb_probe':imacfb.c:(.init.text+0x1039): undefined reference to `efi'
[10:08] <infinity> That first one definitely shouldn't be happening...
[10:09] <alex_joni> I see it still is included by a #define CONFIG_X86_IO_APIC
[10:09] <infinity> #ifdef CONFIG_X86_IO_APIC
[10:10] <infinity>         check_acpi_pci();       /* Checks more than just ACPI actually */
[10:10] <infinity> #endif
[10:10] <infinity> Yeah.
[10:10] <alex_joni> there's a comment there "Checks more than just ACPI actually"
[10:10] <infinity> You beat me there.
[10:10] <alex_joni> right.. that one
[10:10] <alex_joni> I tried unsetting the X86_IO_APIC, but make oldconfig probably put it back
[10:11] <infinity> I'm assuming that should be #if defined(CONFIG_X86_IO_APIC) && defined(CONFIG_ACPI)
[10:11] <infinity> Or the former should depend on the latter.
[10:11] <alex_joni> I can change that (for a quick fix now)
[10:11] <alex_joni> any idea about the efi?
[10:11] <infinity> Haven't played with EFI on x86 at all yet.  Someone needs to send me a MacTel to play with.
[10:12] <alex_joni> what's efi actually? 
[10:12] <infinity> Early load firmware for ia64 (and now MacTel) machines.
[10:13] <alex_joni> ok, the acpi stuff now doesn't complain anymore..
[10:14] <infinity> Comparable to, say, OpenFirmware on PowerPC, OBP on Sparc, SRM on Alpha, or the godforsaken PC BIOS on traditional x86 kit.
[10:14] <alex_joni> oh. ok :)
[10:14] <infinity> (Not really comparable to the PC BIOS at all, but for the purpose of booting, serves the same needs)
[10:14] <alex_joni> I see the error comes from drivers/video/imacfb
[10:15] <alex_joni> I'll unset CONFIG_FB_IMAC
[10:16] <alex_joni> seems to have done the trick..
[10:16] <alex_joni> brb, getting some dinner while it compiles
[10:16] <alex_joni> thanks again for all your help & insight