14:00:41 <hhorak> #startmeeting  State of the Fedora Kernel
14:00:41 <zodbot> Meeting started Wed Aug  6 14:00:41 2014 UTC.  The chair is hhorak. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:42 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:01:26 <hhorak> talker: Josh Boyer
14:01:57 <hhorak> request: common base kernel for all stable releases
14:02:18 <hhorak> that means shared bugs, fexes
14:02:36 <hhorak> rebases to latest upstream will be more frequent
14:04:14 <hhorak> rebase flow currently looks quite complicated
14:05:39 <hhorak> currentlu every stable and f21 includes different version, also different from the time the release was GA
14:06:21 <hhorak> #topic State of the Fedora Kernel
14:07:22 <hhorak> currently kernel in bodhi does not get any karma in F19
14:07:38 <hhorak> but maintainers need feedback
14:08:26 <hhorak> supported open releases last year: 854 bugs
14:08:57 <hhorak> this year kernel got only 533 (applause)
14:09:26 <hhorak> what we do now is bug triage -- but it is not work we that makes things done
14:09:54 <hhorak> problem areas: wifi, suspend/resume, video and brightness, platform drivers, bluetooth
14:12:17 <hhorak> chart: bug counts -- total bugs number lowers, new bug number is constant
14:13:42 <hhorak> typical scenario -- needinfo on reporter + "please, test the latest kernel" and nobody replies in two weeks, then maintainers close the bug in two weeks -- that's why number of open bugs lowers
14:14:34 <hhorak> Q: do bugs get reported to upstream?
14:14:46 <hhorak> A: yes, in case there were not fixed there yet
14:15:18 <hhorak> what does this all mean?
14:15:43 <hhorak> rebases help and hurt in fairly equal amounts
14:15:55 <hhorak> new fedora releases effectively kill old ones
14:16:14 <hhorak> kernel maintainers need help
14:16:58 <hhorak> after last flock only some people started to provide more information and searching commit on their own
14:19:14 <hhorak> bugs with patches get included pretty fast
14:19:44 <hhorak> abrt reports are quite unuseful; on the other hand retrace server is better
14:20:53 <sgallagh> hhorak: Clarifying: he said that early ABRT was not useful. He noted that it improved massively and now he prefers ABRT bugs over hand-reported ones because they don't need to keep asking for additional information.
14:21:03 <hhorak> coverity (static analysis tool) is also handy tool
14:21:25 <hhorak> sgallagh: thanks for correcting
14:21:30 <sgallagh> any time :)
14:22:11 <hhorak> f21 will get 3.16 kernel, maybe 3.17, depending on GA date
14:24:48 <hhorak> f21 changes in packaging: auto-provides, compressed modules, kernel-core/kernel-modules split
14:25:52 <hhorak> extras package includes modules that only few people use, but still cannot be removed
14:28:23 <hhorak> compression of modules by xz; loading/unloading performance diff is not big deal
14:29:52 <hhorak> f21: arm64 stuff -- new arm architecture aarch64
14:31:58 <hhorak> aarch64 is trying be ready for the moment fesco comes with "aarch64 is primary architecture" request
14:34:32 <hhorak> playground -- copr build that should work, but provides no guarantee
14:35:19 <hhorak> playground kernel tracks f21/rawhide
14:36:07 <hhorak> playground includes kdbus, overlay v23; other possibilities are kpatch/kgraft
14:39:39 <hhorak> Q: scratch builds get removed, is it possible to provide src rpm for bug somehow?
14:39:57 <hhorak> A: we should take that into account
14:41:20 <hhorak> Q: bug backlog lowers down, should not we focus rather on new bugs?
14:42:00 <hhorak> A: yes, we incoming bugs get higher priority
14:42:37 <hhorak> Q: does rebasing bring unstability?
14:43:04 <hhorak> A: yes and no; rebases help and hurt
14:43:29 <hhorak> Q: could we predict bugs based on upstream?
14:43:46 <hhorak> A: yes, if we have users running rawhide, which is not the case
14:48:29 <hhorak> #endmeeting