[08:33] <maco> discussing kernel bugs goes in the "kernel development discussion" category, right?
[08:37] <soren> Usually.
[08:38] <maco> so how should i go about debugging an issue with a hotkey that apparently is kernel-level because input_events and acpi_listen both say nothing's going on when i press the key?
[09:37] <mnemo> there is something weird happening with the kernel source package... I get this --> http://rafb.net/p/hSVkME70.html
[09:37] <mnemo> but some other people are not getting that problem
[09:37] <mnemo> (im using "main server" in apt software sources config)
[09:43] <TheMuso> mnemo: You are better off fetching the linux source either from git, or with linux=2.6.28-7.20 for example.
[09:44] <mnemo> but the git might contain some patches that are not in my binary right? because I assume a new package is generated automatically every time someting is commited to the git tree?
[09:45] <mnemo> TheMuso: what you mean exactly when you say "use linux=2.6.28-7.20" ??
[09:45] <TheMuso> mnemo: You use the source package name linux, and you are giving an exact version number, so apt-get doesn't get confused and look up the first partial match it finds.
[09:46] <TheMuso> mnemo: I don't know how the kernel team do PPA uploads, but what is in git now may not be in the binaries that are in the archive.
[09:46] <TheMuso> More to the point, I don't know how often they upload to PPAs etc.
[09:46] <mnemo> ok, makes sense
[09:47] <mnemo> but what would be the full apt-get command line for doing the linux=2.6.28-7.20 trick?
[09:47] <TheMuso> apt-get source linux=2.6.28-7.20 as of the latest upload from the kernel team in the archive.
[09:48] <mnemo> how did you find that number? I mean if I want to do this again later on
[09:48] <TheMuso> mnemo: Best place to check is launchpad, http://launchpad.net/ubuntu/+source/linux
[09:51] <mnemo> ok thanks
[09:51] <TheMuso> You're welcome.
[09:55]  * TheMuso sighs. Ok, that should be the final ports upload for now, and should get us binaries on most if not all ports architectures.
[09:56] <TheMuso> All arches would be nice, but there may yet be unknown FTBF issues on ia64, sparc, or hppa.
[11:44] <persia> I'm working on the Jaunty LPIA alternate CD, and have run into issues with not being able to load the CD-ROM driver after first boot under KVM.  It was suggested that this may be a missing kernel module in the udebs, but I'm not seeing much variance in the set of modules built between the regular kernel and the lpia kernel in the debian/d-i/modules directory.  Could someone help me track down this issue, or redirect me to the appropriate foru
[11:44] <persia> m?
[12:58] <johan> What can I do to get a vmlinux file for my standard ubuntu kernel?
[12:58] <johan> Do I have to rebuild the kernel and extract that file from the source tree? Or is there a way to extract a vmlinux out of a vmlinuz?
[13:13] <soren> johan: It's just a gzipped vmlinux, I believe.
[13:15] <johan> soren: google tells me there are no known tools to extract it
[13:15] <soren> gzip?
[13:15] <johan> I'd guess that the symbols are also stripped
[13:15] <rtg> johan: there are still enough symbols to load modules.
[13:16] <johan> rtg: yes, but not enough to get useful information out of oprofile
[13:17] <rtg> other then the exported public symbols, then yes you are correct
[13:17] <johan> there used to be linux-image-debug kernels just for that, but appears they're not updated any longer
[13:18] <rtg> johan: IIRC, Hardy is the last release that produced debug kernels
[13:19] <johan> rtg: is there a reason they can't be built and put on ddeb.ubuntu.com for instance?
[13:20] <rtg> I think we determined that it was a lot of build/space overhead for little benefit. no one seemed to be using them.
[13:23] <johan> it's pretty useful to have them for tools like systemtap/oprofile
[15:29] <Keybuk> nyargh!
[15:29] <Keybuk> the kernel's silly fd interfaces never cease to annoy me
[15:30] <Keybuk> they're implemented as streams, yet don't behave like streams
[15:30] <Keybuk> you can't just read part of the data into a buffer, it returns -EINVAL if the buffer is too small for the struct
[15:31] <Keybuk> but you can over-read multiple structs in one read
[15:31] <Keybuk> should so be a msgbuf instead
[21:18] <Kano> hi rtg 
[21:23] <rtg> Kano: you're on my todo list for tomorrow
[21:23]  * laga blinks
[21:23] <Kano> thats a tiny patch against the driver you already have, i tested it as external compile hack, it compiled
[21:23] <Kano> but the user with the webcam did not appear, all i can say it compiled
[21:24] <rtg> Kano: if it doesn't fix anything, then why bother
[21:24] <Kano> it fixes the compilation,the tiny patch i gave you in the last mail
[21:24] <Kano> i guess it could work
[21:25] <Kano> it was always the same abi change
[21:25] <Kano> so i guess the patch is correct