fedora-kernel
LOGS
18:00:49 <jwb> #startmeeting
18:00:49 <zodbot> Meeting started Fri Nov  2 18:00:49 2012 UTC.  The chair is jwb. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:00:49 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
18:01:01 <jwb> #meetingname fedora-kernel
18:01:01 <zodbot> The meeting name has been set to 'fedora-kernel'
18:01:10 <jwb> #meetingtopic Fedora Kernel
18:01:15 <jwb> #topic init
18:01:24 <jwb> should be a quick meeting today.  who's here?
18:01:32 * nirik is lurking around
18:01:34 * jforbes is here
18:01:49 <pknirsch> jsmith: thanks! :)
18:01:52 * brunowolff is here
18:02:07 <jwb> davej, ?
18:02:25 * jsmith lurks just for kicks and giggles
18:03:02 <davej> sorry, here. (dude fitting new storm doors has the best timing apparently)
18:03:27 <jwb> should we do the release overview quickly?
18:03:40 <jwb> #topic F16
18:03:57 <jforbes> F16 has a 3.6.5 update waiting for your karma
18:04:24 <jwb> that pretty much sums it up, doesn't it?
18:04:27 <jforbes> There are a large number of important fixes since this update covers 3.6.2-3.6.5, so the sooner we get it out the better
18:04:46 <davej> I needinfo'd a ton of 16 bugs earlier this week. a bunch of them got replied to, some closed because of fixes, some became 17 bugs, but a lot are still problems apparently.
18:04:57 <jforbes> Yeah, bugs are kind of hard to guage at the moment, a lot of the F16 bugs are still waiting for a response from davej's mass update
18:05:06 <jwb> #info F16 3.6.5 in updates-testing covers 3.6.2-3.6.5.  lots of fixes.  please test and add karma
18:06:32 <jwb> ok, moving on
18:06:36 <jwb> #topic F17
18:06:49 <davej> 17 is a bit of a mess right now. Over 400 open bugs.
18:07:05 <davej> I'm getting the feeling a lot of them are related however.
18:07:23 <jwb> yeah.  i had it down to 300 at one point, but the rebase of f16 and other things seemed to have bumped the count back up
18:07:31 <davej> so hopefully some point real soon things will turn around there.
18:08:08 <nirik> seems like there's a number of 'my screen in blank' type video bugs...
18:08:18 <nirik> at least we see a lot of those in #fedora
18:08:24 <davej> during boot?
18:08:25 <jwb> #info F17 3.6.5 in updates-testing.  please test and give karma
18:08:55 <jwb> nirik, the ones we've been calling "fb bugs"?
18:09:03 <nirik> davej: yeah. Usually it's "I updated and rebooted and got a blank screen"... but it's hard to pin down what kind of hardware or whatnot
18:09:06 <nirik> yeah
18:09:17 <davej> I think that framebuffer hand-off problem still isn't fixed. Though I don't think we've seen (m)any f16 reports yet
18:09:52 <davej> one day maybe one of us will get hardware affected by it. Until then, that one's a mystery
18:10:18 <jwb> airlied thinks it's not in the fb layer anymore, but in the individual video drivers
18:10:31 <jwb> but we've seen it on i915, nouveau, and radeon so...
18:10:41 <nirik> so (to bring up another topic)...
18:10:50 <jwb> either fb or drm or every driver has a bug that's producing similar results
18:10:50 <nirik> would it be helpful to ask them to try the vanilla kernel?
18:10:59 <nirik> (that thl has been working on providing)
18:11:24 <davej> I'd be surprised if that helped if they're built with the same options
18:11:26 <jwb> possibly.  i'd be surprised if it were a patch we're carrying though.  we aren't carrying any drm/fb patches that i'm aware of
18:11:44 <nirik> ok.
18:12:00 <jwb> #info a number of 'blank screen/hung' bugs reported.  no clear cause yet.  git bisect is welcome
18:12:31 <jwb> anything else on f17?
18:12:31 <davej> previous attempts at bisects haven't worked out so well. not sure what to make of that
18:12:38 <jwb> davej, yeah...
18:12:43 <jforbes> It's a race, and people have missed it just rebuilding locally
18:12:49 <davej> the closest we got was the merge commit, which...
18:13:30 <davej> it might need something more intrusive like early console debugging
18:14:00 <jforbes> davej: the first part airlied found required hacking netconsole to be able to see anything at all
18:14:14 <davej> yeah, it's messy debugging stuff that early
18:15:07 <davej> think that's it for 17 ?
18:15:17 <jwb> think so
18:15:21 <jwb> #topic F18
18:15:41 <jwb> last night adamw and the QA team got 3.6.5-2.f18 approved for the beta
18:16:13 <jwb> that contains a backport of the modsign code in 3.7-rc2 and an update to the secure-boot patchset to import certificates from UEFI in SB mode
18:16:21 <jwb> if you aren't in SB mode, it doesn't even bother
18:16:44 <jwb> there's some small fallout with building external modules against that, but i hope that will be resolved soon
18:16:49 <jwb> i have a patch sent to rusty to fix it
18:17:08 <jwb> we'll have at least one more kernel build needed for beta
18:17:28 <jwb> it will probably be 3.6.6 plus some small changes
18:17:40 <brunowolff> Are you going to wait for the patch to be accepted upstream before including it in fedora kernels?
18:18:03 <jwb> brunowolff, i'd like to, yes.  hopefully rusty pulls it in tomorrow.  he took the other one already
18:18:13 <jwb> brunowolff, i won't wait for it to get to linus' tree.  just into rusty's
18:18:32 <brunowolff> I'll keep an eye out for it.
18:18:52 <jwb> once i have that in place, i plan on writing up a brief tutorial on how to sign 3rd party modules
18:19:00 <jwb> but unless you are in SB mode, you really don't need to
18:19:39 <jwb> i still need to go through the open bugs, but overall i think we're in decent shape.
18:19:45 <davej> worth noting that we've got around 50 f18 bugs open, vs the 400 f17 bugs, and they're the same kernel.  F18 still not getting a huge amount of testing. hopefully that will change with the beta.
18:20:08 <jwb> right.  was just going to say "of course, we still have all the F17 bugs probably lurking"
18:20:37 <davej> we should probably be prepared for a bunch of the 'blank screen' bugs when beta lands
18:21:13 <jwb> yeah :\
18:21:28 <jwb> any other comments/questions on f18?
18:21:41 <davej> I'd *really* like to get a handle on the webcam thing too. Going to see if I can devote some time to that next I think
18:21:50 <jwb> #info 3.6.5-2.fc18 is the current kernel.  one more kernel update coming before beta
18:21:53 <brunowolff> If 3.7 is released before final freeze, would you use 3.7 for final?
18:22:00 <jwb> brunowolff, doubtful
18:22:17 <jwb> we've discussed that a couple times.  unless f18 slips by a huge amount, we'll probably just do an update for 3.7
18:22:40 <jforbes> brunowolff: we had discussed last time, it might be a 0 day update, but we would rather have more time testing since media can't be respun
18:22:46 <jwb> #info F18 GA kernel level is still expected to be 3.6.x.  3.7 will be a rebase update
18:23:51 <brunowolff> That's kind of why I asked. I would switch to using 3.7 everywhere if it looked like 3.7 might be used for the release. Otherwise I just use it for my rawhide machine for now.
18:24:15 <jwb> ah
18:24:24 <jwb> oh, we should talk about update paths maybe
18:24:44 <jwb> the QA team expressed frustration that f17 had a newer kernel out there than f18
18:24:56 <jwb> which is partly why 3.6.5-2.fc18 got accepted during freeze
18:25:10 <jwb> i'll try and keep this in mind going forward, but i don't see a magic solution here
18:25:19 <jwb> we can't sit on updates for the majority of fedora users
18:25:49 <davej> especially when those updates are security fixes, unless we start munging version numbers (no)
18:26:15 <jforbes> Well, it really only impacts upgrades right?
18:26:17 <brunowolff> If you think there might be a version issue, could you use the fc17.1 type naming or is that especially painful for the kernel?
18:26:30 <jwb> jforbes, yeah
18:26:41 <davej> I think as long as we get things fixed up quickly enough, it's something we'll have to live with
18:26:51 <jwb> it'll be an issue even post-GA
18:26:52 <davej> the freeze process is painful: film at 11
18:27:11 <jforbes> so the testers could easily remove the newer kernel. before the test.
18:27:19 <jwb> that too
18:27:36 <jwb> except the current instructions say to make sure you update your f17 system first
18:28:12 <jwb> ok, moving on
18:28:17 <jwb> #topic rawhide
18:28:29 <jwb> rawhide is chugging along.  currently at 3.7-rc3.git4
18:28:38 <jwb> it continues to get little testing
18:29:00 <jwb> as expected, 3.7-rcX is looking good on my machines at this point, but we need more testers
18:29:48 <jforbes> Maybe it is time to start the rawhide-nodebug kernel repo
18:29:51 <jwb> #info rawhide is at 3.7-rc3.gitX.  will continue through the 3.7 RC releases
18:29:56 <brunowolff> As soon as I can build dahdi-linux again, I'll be using 3.7 kernels on my main desktop at home.
18:30:16 <jwb> jforbes, possibly!  something i was thinking of doing once i get f18 in order
18:30:31 <jforbes> I have repo space already set up, I can start doing that, but should we build it against the current stable release or against alpha/beta?
18:30:48 <jwb> #idea ramp up the 'rawhide-nodebug kernel repo'
18:30:59 <jwb> jforbes, why wouldn't we build it in koji via scratch?
18:31:38 <jforbes> jwb: we would, but we still should probably use a build target of a release people will install it on
18:31:54 <jforbes> jwb: Oh, NM.  has to be mock
18:32:13 <jforbes> jwb: since we don't want to commit turning off debug every day, then turning it back on
18:32:33 <jwb> right.  or scratch with --srpm
18:32:34 <jforbes> Bah... NM. typing before thinking through, srpm push to koji
18:32:49 <jwb> i just think of koji scratch as "mock not on my machine"
18:33:21 <jforbes> So still, we need to pick a target. I would probably recommend newest stable, since we don't want to take F18 testers away from F18, there are so few as it is
18:33:44 <jwb> i'm still confused
18:33:54 <jwb> what target?  just scratch build it for rawhide
18:34:34 <jforbes> 99% of the time that's fine
18:34:48 <jwb> cases where it isn't?
18:34:52 <jforbes> Actually, we have to anyway.
18:34:53 * nirik is considering making a push to use rawhide day to day and get others to in the f19 cycle...
18:35:08 <jforbes> Occasionally other bits of rawhide are broken and cause issues.
18:35:28 <jforbes> But more frequently we have buildreqs which are not in current stable releases (pesign, etc)
18:36:00 <jwb> yeah.  and in the case of "f19 compiler broke something" we want to know that asap anyway
18:36:18 <jwb> where by broke i mean "miscompiled" not "failed build"
18:36:27 <jwb> nirik, i plan on running a rawhide machine during f19 too
18:36:41 <jforbes> Okay, so will start building against rawhide. How many kernels do we want to keep in the repo? just latest?
18:36:59 <nirik> cool. I was thinking of making a group of known rawhide day to day users... could help each other and fix stupid things that get broken.
18:37:17 <jwb> jforbes, probably to start with
18:37:38 <jwb> it would be nice if we could see how much usage that repo is getting too, but i'm not sure how to accomplish that
18:37:38 <jforbes> can quickly get huge if not, so we would need a really good reason to keep more
18:37:39 <davej> nirik: it would be nice to have more organisation about rawhide, so people are aware of what's busted from day to day.
18:37:59 <davej> it's no fun updating, finding shit's all fucked up, and then reporting and finding out that a bunch of people already knew
18:38:04 <jforbes> jwb: I can see if we can get request counts
18:38:08 <nirik> yeah, also more provenpackagers using it day to day to go fix junk that happens as it happens...
18:38:27 <nirik> instead of weeks later
18:38:52 <jwb> jforbes, yeah, cool
18:39:13 <nirik> anyhow, beyond the scope here, just thought I would mention it.
18:39:19 <brunowolff> It would be nice to get updates-testing packages from branched (while there is a branched) into rawhide as well. As otherwise stuff stays broken there longer than in branched on a regular basis.
18:39:19 <davej> perhaps even a #fedora-rawhide channel would be useful. it's too easy to miss stuff on -devel
18:39:38 * nirik nods.
18:39:51 <nirik> and rawhide users are not all developers.
18:40:06 <nirik> brunowolff: I would prefer to not have any inheritance at all.
18:40:09 <davej> anyway, kind of deviating away from kernel stuff here...
18:40:16 <nirik> but we have had that discussion before.
18:40:24 <jwb> #info rawhide needs love.  much love.
18:40:38 <jwb> ok, let's move on to open topics
18:40:42 <jwb> #topic open floor
18:41:06 <brunowolff> If the developers did rawhide builds that would be fine, but in practice a lot don't want to do extra builds for rawhide.
18:43:22 <nirik> brunowolff: IMHO they should suck it up... but thats just me... so I will shut up now. ;)
18:43:26 <jwb> any open floor topics?
18:45:15 <jwb> i know knurd has a number of RFCs out on the kernel list.  i think we're still working through some of them
18:45:43 <jwb> and i don't think he's present today
18:45:51 <jwb> something about vacation.  whatever that is
18:46:15 <jwb> ok, let's call it a meeting.
18:46:21 <davej> done
18:46:22 <brunowolff> Thanks
18:46:28 <jwb> thanks for attending everyone
18:46:31 <jwb> #endmeeting