18:01:11 <jforbes> #startmeeting kernel
18:01:11 <zodbot> Meeting started Fri Sep  6 18:01:11 2013 UTC.  The chair is jforbes. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:01:11 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
18:01:27 <jwb> yay
18:01:30 <jforbes> #topic who's here?
18:01:38 <davej> I am bringing the noise.
18:01:50 <jforbes> does that mean jwb's got the funk?
18:02:14 <jforbes> #topic F18/F19
18:02:43 <jforbes> Okay, I will cover this one.  3.10.11 should release over the weekend. Both F18 and F19 will get that.
18:03:08 <jforbes> Shortly afterwards, F19 will get the 3.11 rebase (people have been asking and seem to be looking forward to it)
18:03:38 <jforbes> perhaps a week later, F18 will get the 3.11 rebase.
18:05:14 <jforbes> As we discussed in the last meeting, the karma attention we have been getting is great, but it meant that kernels were often not even hitting updates-testing.  With the 3.10.10 update, autokarma was disabled until the update had been in the updates-testing repository for 24 hours, and then turned back on.  This worked out very well, so we will be doing that in the future as well.
18:05:36 <jforbes> The 3.11 rebase will have a slightly longer period in updates-testing just to ensure it gets coverage
18:06:00 <jforbes> Otherwise, the mei issues seem to be in much better shape.
18:06:30 <jforbes> #info F19 will get 3.10.11 before a rebase to 3.11 kernels
18:06:33 <jwb> with the exception of people who can't even load it without it causing issues
18:06:44 <jwb> blacklist is their only option
18:07:12 <jforbes> #info F18 will get a 3.11 rebase shortly after F19 (a week or so)
18:07:41 <jforbes> right.  Hopefully that can be figured out soon.
18:07:49 <jforbes> Anyone have anything else on F18/F19?
18:08:17 <davej> nope
18:08:28 <jwb> i'll be doing mass bug needinfos when the rebase goes in
18:08:36 <jforbes> #topic F20
18:08:49 <jwb> so F20 is sitting at 3.11.0 right now
18:09:04 <jwb> we'll be migrating along the 3.11.y releases likely up through GA
18:09:29 <jwb> the biggest change otherwise has been the switch of our secureboot support.  we've rebased onto a secure_modules patchset by mjg59
18:09:42 <jwb> and dropped some functionality that nobody was using and wasn't going to be accepted upstream
18:10:06 <jforbes> (that change will not propogate down to F18/F19 with the rebase)
18:10:17 <jwb> bugs coming in have been a slow trickle so far, but i think they'll pick up when Alpha goes out the door
18:11:10 <jwb> i can't think of much else for F20 at the moment
18:11:25 <jforbes> #topic rawhide
18:11:40 <brunowolff> The probelm with the default kernel being set correctly for PAE kernels has been fixed.
18:11:45 <jwb> rawhide is rolling along the 3.12 merge window kernels as of right now
18:12:11 <jwb> thus far we've had USB, drm, tty, driver-core, networking, fs, and a ton of other subsystems merged
18:12:40 <jwb> i've been testing them on every machine i have and i haven't hit any issues thus far.  that isn't saying tons, but at least they're known to work on a variety of machines
18:13:09 <jwb> at some point, i expect we'll start looking at bringing in the secureboot kdump patchset posted yesterday
18:13:16 <brunowolff> I running 3.12 kernels on two machines since last night and haven't seen anything odd.
18:13:19 <jwb> but that still needs some revisions
18:13:27 * nirik hit a backlight issue this morning...
18:14:04 <jwb> nirik, file a bug?
18:14:25 <nirik> yeah, I will if it happens again... updated and rebooted and haven't had a chance to see if it happens still.
18:14:26 <jwb> using koji builds to bisect might be kind of hard.  ARM keeps failing builds
18:14:39 <jwb> and perf
18:15:45 <jforbes> anything else on rawhide?
18:15:51 <jwb> i can't think of anything
18:16:00 <jforbes> #topic coverity/trinity/upstream
18:16:07 <jforbes> davej: want to chime in on this one?
18:16:14 <davej> sure
18:16:32 <davej> since the merge window opened, I've been doing daily builds with coverity
18:16:36 <davej> it's been.. interesting.
18:16:57 <davej> day 1 had over 400 new issues, but subsequent days less so
18:17:07 <davej> I put this down to drivers/staging being in the day 1 merges
18:17:19 <davej> who knew that was awful code ?
18:17:54 <davej> as well as doing the builds, I've been picking through their database coming up with patches, and sending mails to maintainers where I wasn't sure if something was real or not
18:18:05 <davej> and a lot of time just marking things as false positives/intentional
18:18:14 <davej> so that's taken up most of this week
18:18:32 <davej> I spent a day reworking how I do those builds, and now it's a few hours faster
18:19:07 <davej> takes about 5hrs from start to finish, and is pretty much automated now apart from me pressing 'go' and babysitting it
18:19:27 <jwb> good
18:19:31 <davej> trinity stuff: pretty quiet so far this merge window
18:19:33 <jwb> what made it a few hours faster?
18:19:50 <davej> jwb: running it on a 24 way machine instead of a 4-way ;)
18:19:56 <jwb> hahaha
18:20:03 <davej> and also upload from the lab->phx is faster than my cable
18:20:11 <davej> well, most the time
18:20:22 <jforbes> nice
18:20:30 <jforbes> Anything else?
18:20:45 <davej> still chasing down some stuff with DEBUG_KOBJECT
18:21:07 <davej> that new DEBUG_KOBJECT_RELASE thing looks like it finds actual bugs, but it's a nightmare to get traces out because it blows up so early on
18:21:25 <davej> that's been my week.
18:21:39 <jforbes> #topic Testing
18:22:33 <jforbes> On the way to getting the regression test suite packaged.  I have filed a request for a badge.  Though who knows how long that will take.  Since there is now a badge for submitting new badge ideas, it seems the queue is long
18:23:13 <jforbes> One thing I did find out is that we can't just throw a message on the fedmsg bus, we are going to have to basically use a small http service as a broker
18:24:32 <jforbes> The good news is, it can also give us some reporting.  So the result of a test run will report who ran it, what kernel version, arch, and pass/fail.  If fail, it will note which specific tests failed. This will all be opt in of course, the test suite will run without any of it, but it is required for badge participation
18:25:08 <jwb> jforbes, can we also use that directly, or hosted on the same server elsewhere, to post the results of the automated test harness runs?
18:25:42 <jforbes> I think that collecting as little data as we are, there shouldn't be any concerns over it.  I will also make it optional for the user bit.  We can't award badges without user, but some people might not care about that, but still be willing to post results
18:26:20 <jforbes> jwb: yes, the auto harness will report as well, using a username we can easily track it on.
18:26:34 <jwb> nice
18:26:57 <jforbes> jwb: and a webpage with a nice little graph so we can always see its results, or we can look at a more broad view
18:27:26 <jforbes> #topic open floor
18:27:41 <jforbes> Anything else from anyone?  Will leave the floor open for a couple of minutes
18:28:26 <jwb> i think that covered everything we said we were doing from the last meeting
18:28:40 <jforbes> #info F20 should remain on 3.11.x until release
18:29:12 <jwb> we'll likely cancel the next meeting
18:29:16 <brunowolff> Does Philip Loughers LZ4 patches for squashfs have a chance of going forward for 3.12?
18:29:19 <jwb> 2/3 of us will be at LPC
18:29:22 <jforbes> Oh, right
18:29:43 <jwb> brunowolff, i've no idea.
18:29:45 <jforbes> #info September 20 meeting cancelled as 2/3 of the kernel team will be at LPC
18:30:41 <jforbes> Okay, on that note, thanks for coming out.  As always, we are around in #fedora-kernel
18:30:45 <jforbes> #endmeeting