UbuntuIRC / 2009 /08 /14 /#ubuntu-kernel.txt
niansa
Initial commit
4aa5fce
[10:13] <krgn> good morning
[10:14] <krgn> hey, is there a way to remove certain arches from debian/config/* ? updateconfigs fails because I removed some dir's for arches i don't want to build for
[10:17] <Daviey> krgn: Wouldn't it make more sense to only build for the arches you want to build for?
[10:18] <amitk> krgn: if all you need is a single flavour, just using the binary-<flavour> target
[10:18] <krgn> Daviey: true, but when using updateconfigs target i have to go through tons of configs which I am too lazy to do ;)
[10:18] <amitk> fakeroot debian/rules binary->flavour>, that is
[10:18] <Daviey> ^^ Surely you can do that :)
[10:19] <krgn> I guess I can just switch myself off for a little while
[11:10] <krgn> hey, is there a way to get sources for the mainline kernel anywhere?
[11:12] <Kano> git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
[11:12] <Kano> ?
[11:12] <krgn> Kano: ok ok, but I also need the ubuntu-specific stuff
[11:13] <Kano> well you can use a ubuntu kernel then git pull from that
[11:13] <Kano> that will update it
[11:13] <krgn> you mean, git checkout origin/v2.6.29 or something
[11:13] <krgn> ?
[11:14] <Kano> git clone git://kernel.ubuntu.com/ubuntu/ubuntu-karmic.git
[11:14] <Kano> cd ubuntu-karmic.
[11:15] <Kano> git pull git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git master
[11:15] <Kano> for example
[11:16] <krgn> Kano: great, I'll try this. thanks
[11:17] <Kano> i am waiting for next official sync currently but you can do that to update that whenever you want
[11:30] <krgn> hmm weird, I end up without debian/rules etc...
[11:48] <krgn> hey, are there deb source files for the mainline kernel builds?
[11:49] <amitk> krgn: I pointed those out to you yesterday
[11:49] <krgn> amitk: this page: http://kernel.ubuntu.com/~kernel-ppa/mainline/ ?
[11:49] <amitk> krgn: https://wiki.ubuntu.com/KernelTeam/MainlineBuilds?action=show&redirect=KernelMainlineBuilds
[11:50] <amitk> go to the archive link on this page
[11:50] <amitk> krgn: your link should work too
[11:51] <krgn> amitk: ah but these don't have source debs so I can see how the packaging is done
[11:51] <amitk> krgn: http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.31-rc4/linux-source-2.6.31_2.6.31-020631rc4_all.deb <-- this doesnt?
[11:51] <amitk> apw: ^^^
[11:52] <apw> yeah i thought we had source debs in the mainline builds?
[11:53] <krgn> amitk: ah yes, but then I need to use specific version 2.6.29.6 as I patch with rt patch and aufs, and the linux-source deb doesn't contain a lot of the files in debian/*
[11:53] <krgn> its quite convoluted :)
[11:53] <apw> krgn?
[11:53] <apw> oh hrm source .debs ...
[11:53] <apw> are we copying the wrong one?
[11:54] <krgn> apw: you mean wrong deb? I think that .31 would be ok, but I can't use that yet since aufs doesn't have a compat patch for rt
[11:55] <apw> well the way they are made is we simply checkout the mainline version, then checkout the debian directory from the tip of the nearest release and run updateconfigs
[11:55] <apw> then build, there is no rocket science involved
[11:57] <krgn> apw: ah yes, thats exactly what I did. then I ran into build failure that only happens with the debian/rules binary-arch but not when I just put a config and run make
[11:57] <krgn> which is strange to me, since it doen't seem to patch anything
[11:58] <apw> what was the build failure as we don't mod the source
[11:59] <krgn> it fails to build kernel/power/snapshot.c because apparently there is a parse error, but this doesn't happen when I do the build manually
[12:01] <krgn> http://pastebin.ca/1529303
[12:07] <amitk> apw: what was the rune to quickly create a branch? 'git checkout -b foo' takes a while
[12:07] <apw> if you don't need to be on it
[12:08] <apw> then you can make a branch directly with git branch <name> <commitish>
[12:09] <amitk> aaaaaaaah, I do need to be on it.
[12:16] <apw> if you need to be on it i don't know of a quicker way
=== testz3urfhdj is now known as ukev
=== ScottK2 is now known as ScottK
[13:44] <ukev> hi... did you heared of the linux kernel root exploit?
[13:44] <ukev> I've just tested it, jaunty full patched is vulnerable..
[13:47] <ukev> no one here?
[13:47] <ukev> its important
[13:48] <amitk> ukev: our -security teams are aware of it and kernel updates are being prepared
[13:49] <ukev> ok great :)....
[13:49] <ukev> thanks for all the work!
[13:49] <amitk> ukev: you could join #ubuntu-security to track status
[13:49] <ukev> ok thanks for that hint
[13:58] * krgn wonders... are any .h / .c files by default in debian/ dir?
[13:58] <krgn> this build error could be related to mismatching source/header files
[13:59] <krgn> for example, there are 3 different files called power.h in debian/ while this also exists in kernel/power/
[15:22] <krgn> sorry for all teh confusing and ... thanks for the help!
[15:58] <amitk> usb oops and apport == fail (atleast in my case)
[15:59] <amitk> lsusb was stuck in a loop and hence apport for stuck
[15:59] <amitk> s/for/was
[17:23] <krgn> is this a common mistake?
[17:23] <krgn> EE: Previous or current ABI file missing!
[17:24] <krgn> I built binary-generic target, and so for this kernel version only one abi exists
[18:08] <jnfuller> can I ask a kernel compilation question here or is this only for development discussion?
[18:16] <jnfuller> mostly what I am wondering is: if I am currently running 2.6.28-14-generic (#47) and I apt-get linux source I end up with 2-6-28-9. If I apply Linus' patch for the null pointer exploit to this tree and then make and install the deb for the custom kernel what sort of issues will I have in relation to the restricted-modules. Will I need to recompile those for the new custom kernel?
[18:17] <jnfuller> Or... is the patch in the works for a new kernel from the offical repos and I shouldn't bother?
[18:21] <rtg> jnfuller, the latter
[18:28] <jnfuller> thanks, rtg, but if I do end up having to apply this kernel (i.e. boss demands it on production systems) I'd have to apply the kernel, reboot and then compile the restricted-modules against the custom kernel in order for the proprietary stuff to work, right?
[18:30] <rtg> jnfuller, any time you change the ABI, then you'll have to recompile all external, 3rd party modules.
[18:30] <jnfuller> ok that's what I thought.
[18:30] <jnfuller> thanks again
[18:33] <krgn> hey, how do I bump the abi? do I do this in the changelog?