/srv/irclogs.ubuntu.com/2016/06/21/#ubuntu-kernel.txt

=== cpaelzer_ is now known as cpaelzer
xnoxapw, right and last week i was like "somebody tell me if this fixes it for them or not"08:38
xnox=)08:38
apwxnox, this fixes it for me :)08:50
xnox=)08:52
ricotzapw, hi, you might be interested in https://paste.debian.net/plain/74070311:38
ricotz(full log at https://launchpadlibrarian.net/266795156/buildlog_ubuntu-yakkety-amd64.firefox_48.0~b2+build1-0ubuntu0.16.10.1_BUILDING.txt.gz)11:38
ricotzrtg, hi, or you ;) ^11:41
apwricotz, i think thats something that broke in a stable update, and i believe we are respinning at the moment for this12:08
ricotzapw, I see, the xenial build seems to work while it uses 4.4.0-24.43 instead of 4.4.0-25.44 12:10
ricotzapw, I guess this only affects x86(_64)?12:11
apwricotz, right ... the fix for a cve introduced the issue, and we have a fix on deck for that i believe12:11
ricotzapw, is there an eta or a bug report to follow?12:12
apwhttps://bugs.launchpad.net/ubuntu/+source/linux/+bug/159293012:14
ubot5Launchpad bug 1592930 in linux (Ubuntu) "failures building userspace packages that include ethtool.h" [Medium,Triaged]12:14
apwricotz, ^ that appears to be the bug we are tracking under12:14
ricotzapw, alright12:14
apwricotz, i believe the only affected version is in xenial-proposed and that one will replaced as soon as the new one builds12:14
apwricotz, and indeed it may alreday be built and pending review to copy, i'll check next12:15
ricotzok, yakkety too of course12:15
apwricotz, yes it gets copied there too12:17
ricotzthanks12:19
=== ratliff_ is now known as ratliff
keeswhen/where are the kernel team meetings these days? it seems https://wiki.ubuntu.com/KernelTeam/Meeting is very out of date :)17:26
apwkees, we don't have scheduled ones any more, noone came for 2 years, so we switched to a newsletter form17:28
apwkees, do feel free to make free with this channel to start discussions any time17:29
keesapw: ah! okay. well, in that case, I'd like to discuss a packaging/building changes for future kernels17:30
keesor rather, give a heads-up.17:30
keesI'm in the process of merging support for gcc-plugins into the upstream kernel17:30
keesthis means a few things:17:30
kees- building the kernel with gcc-plugin-enabled features will require the gcc-N-plugin-dev package added17:30
kees- the kernel-headers package will either need to depend on a new binary package, or will need to become a binary package17:31
kees- said binary package will need to include the .so files that are the built gcc plugins from the kernel build17:31
kees(so that out-of-tree builds can run the plugins)17:31
keesDoes any of this sound terrible, and either way, is there anything I can do to help with it?17:32
apwthe headers are already binary arch/flavour specific, we include some kind of .so in the powerpc ones for the same reason; so that sounds tractible17:32
keesoh, excellent. well that's one problem down. ;)17:32
apwthe first is just new build-depends if i am reading things right17:32
keesright, that should be trivial17:33
keesit was the "carry .so files in the headers package" that seemed to me to be the "worst" :)17:33
apwso overall that sounds like its mostly about knowing what to include in the headers packages17:33
keesbut you're already doing that, so yay17:33
apwi beleieve we include the config binary in them for instance so we have to be binary17:33
keesyeah, I'm going to try to figure out what it looks like for out-of-tree builds. I assume some symlinks or something need to be added/tweaked/etc17:34
apwwhen is all this slated to land ?17:34
apwyeah i assume it all lands linked out of /lib/<version>/build/lib or something17:34
keesthe first plugin should be landing in 4.8. more should follow17:34
apwlib/modules/<version>17:34
apwdamn so we will likley have to figure this out for yakkety then, could you copy me on the patches as and when17:35
keesI think it would end up being /lib/modules/VER/build/scripts/gcc-plugins/ right now, but yeah17:35
keessure thing17:35
apwok so we include scripts already, so it might just happen "naturally" but we clearly need to track17:35
* apw dumps this info in our "things to fear" list17:35
apwkees, and thanks for the heads up17:35
keesyeah, and I think we'll want to first plug ("latent_entropy") enabled by default on x86, arm, and arm64 at least.17:36
keessure thing!17:36
keess/want to/want the/17:36
* apw hopes the patches will make it clear what is going on :)17:38
kees... I'm hoping so ... but ... the kbuild parts are dark magick that I am only now starting to understand.17:38
keesand the plugins themselves are compiler dark magick... but they have a lot of comments.17:39
apwyeah kbuild is deeply arcane black magic, i am supprised it is legal17:39
keeshehe17:39
=== JanC is now known as Guest79667
=== JanC_ is now known as JanC
kees(in related news, can't this spl thing be a pre-built dependency, it takes forever to run its configure script in the kernel build)17:55
cyphermoxapw: hey18:34
cyphermoxdid you have a look at the testing matrix for secure boot? or is it all just for rtg? 18:35
apwcyphermox, i cast an eye over it, i've not got to do anything aboutit yet19:48
cyphermoxapw: fair enough20:09

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!