21:00:19 #startmeeting 21:00:19 Meeting started Wed Jan 9 21:00:19 2013 UTC. The chair is pwhalen. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:00:19 Useful Commands: #action #agreed #halp #info #idea #link #topic. 21:00:19 #chair pwhalen jonmasters bconoboy ctyler pbrobinson dgilmore 21:00:19 Current chairs: bconoboy ctyler dgilmore jonmasters pbrobinson pwhalen 21:00:22 .fas jcapik 21:00:22 jcapik: jcapik 'Jaromír Cápík' 21:00:23 .fas pwhalen 21:00:25 pwhalen: pwhalen 'Paul Whalen' 21:00:38 * jonmasters_ is in 21:00:40 * masta waves 21:00:41 .fas jonmasters 21:00:41 jonmasters_: jcm 'Jon Masters' 21:00:45 .fas dmarlin 21:00:46 dmarlin_: dmarlin 'David A. Marlin' 21:00:49 .fas blc@ 21:00:49 bconoboy: blc '' 21:01:14 its late for pbrobinson, so lets get started 21:01:20 #topic 1) Current Problem packages 21:01:28 there's not really and problem packages, eclipse is still an issue but is being worked upon and I've worked around some of it's deps. texlive is still an issue but we know the problem and the resolution. ruby is fixed. gmp is hacked around but we're still awaiting a new binutils to fix it properly. 21:01:53 the major issue is the newRepos causing failed builds. It seems to have settled a bit in the last day or so and we're getting less failed builds or it seems that way at least but not doubt now I've said that it will get worse again 21:02:29 other than that rawhide is moving forward with constant bashing from me 21:03:16 \o/ 21:03:24 and f18 has 280 missing packages and around 47 "older" packages. I'm looking through those but I don't expect that number to decrease much between now and release for f18. There's some that already have fixes in testing though 21:03:24 .fas chris@tylers.info 21:03:25 ctyler: ctyler 'Chris Tyler' 21:03:43 I think somebody was making a list of package we cna look at 21:03:51 that would be me 21:04:17 I'm going through the list of 280 odd and categorising them into x86 etc 21:04:44 when I get a few spare cycles. I should have the list out next week I would guess 21:04:57 I'd like to try to help fixing some, so when the list is ready I'm going to cherry pick 21:05:36 oh the last main package set on my list to bash on is ghc which I'm working with juhp on 21:05:44 #action pbrobinson to send a list of problem F18 packages to the mailing list next week 21:06:02 #info pbrobinson making a list hitlist of packages, dmarlin to help whack-a-mole on that list 21:06:37 anything else pbrobinson? 21:06:42 there's not that many moles to whack. A lot of them are things like ada and pascal that need proper work 21:06:48 EOL 21:07:21 #topic 2) Plans for the 3.8 kernel 21:07:22 pbrobinson: the "proper" work is somewhat what I mean. If it's moderate engineering effort, I'd like our internal guys to assist 21:08:02 jonmasters_: I'll break all that down in the list I send out and you can slice and dice how ever you think it best fits 21:08:14 cool 21:08:55 for the 3.8 kernel, David is going to start doing a few tests (yes, I am still working on 3.7 highbank, but it's worth proceeding with 3.8 work) 21:09:21 I'd like to know e.g. what works and doesn't across a few boards, as well as start thinking about the variant->multiplatform kernel upgrade path 21:09:48 so the current built kernel is basically a 3.7 config plus new configuration options 21:09:52 imx has been removed 21:10:02 (and we won't miss it really at this point) 21:10:08 You won't. :-) 21:10:10 and maybe post rc3 I'll merge that config into unified 21:10:29 I need to have the fix for the highbank errata issue 21:10:40 and you will, today 21:10:58 and I'd sort of like to have that before I enable any more of the unified platforms so we know where we stand 21:11:33 then we can test highbank/versatile as we know they should then both work, and can cross test a 3.7 21:12:04 ok 21:12:12 from there I plan to enable imx so we can hopefully have some of the new imx6 stuff supported as it appears to be really cheap 21:12:28 I agree that the moment highbank is done on 3.7 we need to start testing upgrades in F18 21:12:35 on the tegra/omap side I would like to see that tested when ever 21:13:03 there's no major changes there as I believe them to be good for 3.7 21:13:14 Can TI Sitara be supported in unified? 21:13:53 ctyler: no, none of the TI platforms are unified yet, can we support it in the omap? Is that like a bbone? 21:14:04 Yeah, bbone 21:14:17 the bbone should be supportable in the omap, I need help testing as I don't have one 21:14:28 Looking like omap derivative, but the straight OMAP options don't cut it 21:14:41 tegra now has tegradrm enabled but we're awaiting the libdrm support. 21:14:45 #action jonmasters to provide highbank errata fix on 3.8/3.7 21:15:06 #action pbrobinson to add additional unified configs (imx, etc) once highbank works 21:15:12 ctyler: no, that's correct but AFAICT the options are enabled to support it as of 3.7 but I need some way/one to test it 21:15:14 I know there's support in the OMAP tree for Sitara, I do not know the state of simultaneous regular OMAP support or eventual multiplatform plans for that 21:15:39 * jonmasters_ suggests someone test, concur with pbrobinson 21:15:54 so I think it would be great to write down some roadmap of what devices will be gettign enabled or disabled in the coming kernel versions. 21:15:55 ok, I'll try and sort out the Sitara-vs-OMAP situation 21:15:56 the TI platforms are preparing for unified support but it won't be until 3.9 or even more likely 3.10 21:16:03 yea, indeed 21:16:05 !yikes! 21:16:33 * ctyler thought they were further along 21:16:39 the problem is they have a lot of complex drivers that need a lot of modification and you can't move them until they're all converted 21:16:45 ctyler: TI also is undergoing a restructuring in the OMAP division that has been broadly reported in the media, so I am not too surprised if this will take time 21:16:48 they are probably the furthest along 21:16:51 tegra should be unified in 3.9 21:17:55 anyway. Also i was looking at various bits as part of 3.8 and we should possibly have support for the XP platforms too with the kernel that is currently building. It will need testing too 21:18:08 so to summarise: 21:18:19 I've forgot, when does samsung get enabled ? 21:18:38 unified kernel currently has highbank and versatile, with possibly mvebu xp support 21:19:06 tegra kernel currently has tegra2 with tegradrm and I'll look towards enabling tegra3 where possible 21:19:27 omap has beagle/panda there and should support bbone as well but the later needs testing 21:19:40 masta: not enough of it upstream yet to be useful 21:19:54 would the miraboard be suitable to test mvebu ? 21:20:06 no idea what that is : 21:20:08 :) 21:20:36 armada soc 21:20:43 needs to be armada xp 21:20:53 armada xp != armada 21:20:59 probably for the 3.9 kernel I'll add a armv7-lpae kernel which will have exynos/omap5 support and what ever other A15 boards 21:22:06 the mvebu should support either XP or 370 AFAICT but it depends on the board. In theory it should support a number of boards if the device tree is supplied.... whether that works or not is still a theory 21:22:18 pbrobinson: we were just chatting with ctyler. I asked him to pick up a Chromebook during FUDCon and get the guys at Seneca to help get UCM, etc. working for us 21:22:59 pbrobinson: I think that's a nice way of sharing out the fun there and then we can get to a point where we eventually have an official Arndale/Chromebook multiplatform kernel 21:23:11 jonmasters_: i have a local UCM already done and mostly working, just need to finish it up and submit for a package review, I dropped it to get f18/rawhide back on the straight and narrow 21:23:40 pbrobinson: sure, but you know, let's spread the love if we can 21:24:06 sue, whatever 21:24:17 good luck getting a chromebook, they are sold out .. 21:24:43 I got one in the UK in about 2 days 21:25:10 i havent seen one for sale int he states since about a week or so before xmas.. 21:25:16 shall we move on to F18 final? 21:25:42 yep 21:25:50 #topic 3) F18 final - What's left? 21:26:18 package cleanups especially eclipse 21:27:08 anything from the f18 beta? pkexec? 21:27:16 audio? 21:27:29 still poking at pkexec 21:28:20 the beta is very good 21:29:07 pbrobinson: you mentioned trying pkexec on a XO, would love to hear about that 21:29:24 dmarlin_ was going to assist with pkexec 21:29:26 I'll be running up beta on everything (beagle/pandaES/trimslice) when I get home from Cairo post Friday 21:29:54 masta: and I'll have a number of XOs so I'll ping you when I'm home to find out what I need to test 21:29:54 I'm concerned that on v5 there is a bigger problem in that gdb segfaults. The problem is not pkexec, the problem is that some DBUS services fail to start because they segfault 21:30:04 so pkexec never gets to handle the authentication 21:30:44 On v5, gdb falls over, a bunch of other stuff does. There is a deeper issue we need to look into. It should not hold the release, but we need to close that down 21:30:50 the auth works, but the actual thing to exec fails 21:31:00 jonmasters_: other than the quick mention of it the other day between you masta and I it's the first I've heard of armv5 issues. It's clearly been widely tested 21:31:10 masta: in my experience, it varies by target actually what happens 21:31:30 jonmasters_: I know on yoru plug it was different than on the v7 boards 21:31:35 pbrobinson: on a guruplug, before the holiday break, I established that gdb would fall over 21:31:56 so there is a v5 issue, it shouldn't hold the release, but we need to get that closed down 21:32:17 so just saying that the "pkexec" issue is a broader issue needing attention 21:32:21 jonmasters_: OK, but then you're the first person to report it then? Or have others? 21:32:58 is it replicable? 21:33:09 jonmasters_ seems to have been able to 21:33:11 I think i have heard it before... 21:34:03 So.... is this a blocker or what? 21:34:22 i have to wipe and reinstall the 18 beta, so I will try to crash it.. :P.. not a blocker, i think it has been around for a long while 21:34:24 jonmasters_: is it possible that some things are being built with v7 flags even though we're specifying v5 21:34:52 masta: can you tell me if PK is linked against js? If so I think I might know what the problem is, I was looking at it earlier this week 21:34:56 the pkexec issue was discovered thanks to it breaking yumex in the panda image 21:35:23 I was goign to fix yumex and got into the weeds with PK 21:35:43 one sec 21:36:05 I mean pkexec linked against js 21:36:38 I'm not seeign anything with the string js 21:37:00 FWIW, I don't think it's just a pkexec issue. Like I said, that's a manifestation, but some other things (esp. on v5) are crashing 21:37:12 the x86 version is linked against libmozjs185.so.1.0 according to rpm -qR 21:37:15 e.g. running "gdb gdb" will result in a segfault 21:37:26 looks like recursion almost 21:37:26 I blame systemd 21:37:41 I do mean to look at it, but dmarlin_ has also "volunteered" to look :) 21:37:48 masta: sush... that's not useful 21:38:29 It'll be a kernel bug or glibc, or a library. But what it is not is SELinux or a ulimit 21:38:36 I did the obvious checks before the holidays 21:38:50 jonmasters_: there's an issue with js being built with the wrong flags on v5. I've got it on my list for this week. I wonder if there's other packages that are detecting the cpu bits like js and building with non specified cflags 21:39:22 pbrobinson: that is a reasonable hypothesis, we do know, for example that packages not using %configure but calling configure directly often do not set the correct options 21:39:27 polkit now uses javascript for some of its rule sets and hence it links with libmozjs185.so.1.0 21:39:45 so I wonder if that's the issue for the pkexec side of things 21:39:51 (that configure example specifically occurred with openmpi but there will be others) 21:40:05 sure, that, and in addition the other crashes on v5 21:40:20 Let's note down these two things, note that people are poking, and move on 21:40:40 I don't consider these blockers for F18 at this stage, but instead release note items with fixes pending 21:40:41 masta: do you have a v5 install you can test against if I hack you up a non optimal js package to test against? 21:41:07 pbrobinson: I can craft myself a v5 panda image 21:41:16 pbrobinson: so that is a yes 21:41:43 oh speaking of release notes, I still think the default password needs to be in the install notes. 21:41:50 masta: might not cause an issue on panda because it can exec v7 code 21:41:56 pbrobinson, I also have a GuruPlug 21:42:06 ya every week theres someone who doesnt know the default password 21:42:10 +1 seano 21:42:17 pbrobinson: I'll use a rpi if I have to 21:42:34 pwhalen: can you add that to the documentation? 21:42:35 seano, default password for F18? 21:42:40 it should be there 21:42:43 OK cool, I'll try and get it to masta and pwhalen in the morning 21:42:55 can we put the default pw's in the channel title? 21:42:59 heh 21:43:02 right in the release notes.. :P 21:43:03 pwhalen: can you install a v5 image on your guruplug and do the test? 21:43:15 pwhalen: you could install a test package from pbrobinson 21:43:19 pbrobinson: ^^^ 21:43:20 can we just put the password in the issue file so it's in their face? 21:43:30 jonmasters_, yessir, hes going to send it tomorrow 21:43:45 the password is in the release notes 21:43:47 pbrobinson: Paul has a GuruPlug now and the JTAG adapter for it 21:43:48 pbrobinson: well, we are hoping they change it. We should probably use firstboot on any systems that have video. 21:43:49 I'll check on the password for F18 21:43:52 cool 21:43:55 i have both too.. 21:44:14 ctyler: if they change it then the issue file isn't a security risk ;-) 21:44:29 #info There are known issues with pkexec failing on v5/v7. On v5 the issue appears to be related to a broader problem with certain programs segfaulting. E.g. "gdb gdb" is a reproducer 21:45:04 #action Paul Whalen to test a pkexec js linkage package fix from pbrobinson, pwhalen, dmarlin and jonmasters to assist in tracking down broader v5 issue 21:45:20 #info Do not consider pkexec issue to be a release blocker at this time 21:45:36 seano, all install pages for F18 Beta have the password, where were you referring to ? 21:45:50 pbrobinson: if you disagree, feel free to undo the last info 21:45:54 so is the pkexec issue just on v5 or is it an issue on v7 too? 21:45:58 I was looking at the release notes on the download page.. 21:46:05 pbrobinson: it's on v7 too, but it's a different manifestation 21:46:19 if it's just v5 I don't see it blocking. 21:46:26 what's the difference? 21:47:16 on v7 it always outputs "Terminated" 21:47:38 e.g. when trying something like 'pkexec ls' 21:47:59 will prompt for creds and then emit "Terminated" 21:48:24 according to the logs the auth is working 21:48:26 OK 21:49:18 on to FUDCon? 21:49:35 yay 21:49:38 +1 21:49:39 yep 21:49:41 #topic 4) FUDCon Planning - a) What infrastructure is needed (switches? arm hardware?) 21:49:49 * ianweller feels summoned 21:49:59 A friend finder is always good, apparently 21:50:04 always 21:50:13 power strips 21:50:23 oh ya power bars 21:50:34 just make sure you bring your uk power converter things 21:50:34 and extension cables 21:50:35 an LCD is probably worth having 21:50:41 will there be coffee? WE need to keep jonmasters at full power. 21:50:52 bconoboy: yes to coffee 21:50:59 some connectivity for non wireless devices if we're working on them 21:51:06 bconoboy: I'm staying with someone who has a coffee machine that will be fueling my mornings 21:51:28 i will say right now, we have ordered coffee for all three days of the event for the entirety of the days 21:51:32 I have a 4 port gigabit switch I can bring if needed 21:51:37 pwhalen: laptops can connect via wireless and share to wired 21:51:41 ianweller: do you have wired ethernet connectivity? 21:51:42 ianweller: yay 21:51:45 jonmasters_: no 21:51:48 sorry :( i tried 21:51:49 but KU is picky 21:51:49 do we get google fiber to the curb there? 21:51:51 bconoboy: bring it 21:51:55 ianweller: wireless? 21:51:59 fossjon: i wish we had it even close to lawrence 21:52:00 ianweller: or should we figure out a...ok so a transparent bridging router is needed then 21:52:05 bconoboy: there is wireless. i think it has a captive portal. 21:52:21 * masta will be driving so it goign to bring at least one lcd, some keyboards, mice, and possibly some old switches/cables 21:52:21 we can bridge with a laptop 21:52:31 I'll bring a small switch too 21:52:34 bconoboy: you have a less batshit crazy idea, thanks 21:52:44 we'll NAT our connections thru ctyler's laptop possibly! 21:52:47 * pbrobinson is going to ponder the back of his eye lids if he's not needed for anything else 21:52:48 I'll bring my EEE PC. It will be good for it to be useful again. 21:52:51 masta: thank you sir, we'll appreciate you bringing a monitor if possible for sure 21:53:12 pbrobinson: thanks for joining today! 21:53:33 pbrobinson: thanks and good night 21:53:38 ianweller: do any of the projectors have HDMI or DVD-I inputs? 21:53:48 DVI-D rather 21:53:48 I'll also bring a trimslice with a copy of the latest fedora arm repos 21:53:58 I will bring an MK802 (for giggles), a model B Pi, a Chromebook, a PandaBoard, a GuruPlug, a hardware debugger, and probably some other stuff too 21:54:03 so we'll have a local copy of needful bits 21:54:04 ctyler: i won't have any idea until i can do a full room survey next week 21:54:11 oh yea... a recentish rsync of the repos would be great on some portable 21:54:19 you might need a digital-to-VGA adapter unfortunately 21:54:22 bconoboy: you're on the hook for the repos then? 21:54:22 ok 21:54:31 it's really a 50-50 chance of the projectors having digital inputs in my head 21:54:36 This is what we have available: http://etherpad.proximity.on.ca:9001/p/rPCVMT2ztd - Any requests? 21:54:37 jonmasters: yep 21:54:42 #action bconoboy to bring a recent rsync on the F18/F19 mirrors 21:55:17 btw http://arm.koji.fedoraproject.org/koji/taskinfo?taskID=1359891 is the kernel to test highbank/omap/tegra/mvebu/vexpress against once it finishes building (which won't be long) if you're bored 21:55:25 Oh, pwhalen put me on the hook for discussing LEG a little at FUDCon 21:55:41 +1 on LEG 21:56:28 #action jonmasters_ to discuss Linaro Enterprise Group at FUDCon Lawrence 21:56:39 ok, do we need a wiki page summarizing who is bringing what to FUDCon? 21:56:55 perhaps pwhalen can take the action of poking us to add things to one if he starts it 21:57:03 then we can see if we're missing anything 21:57:13 I'll drop the link in channel 21:57:22 * jonmasters_ is flying this Sat to CA, then on to FUDCon next week, so I'll need to pack myself this week 21:57:46 Where CA is a state or country code? :-) 21:57:57 california, canada 21:58:05 :) 21:58:44 it'd be more fun if I was flying to Ontario, CA 21:58:50 but sadly not this time ;) 21:58:57 #topic 4) FUDCon Planning - b) Attendee arrivals and departures 21:59:19 * jonmasters_ is arriving on Thu night after midnight 21:59:27 * jonmasters_ is departing on Sunday around 3 or 4pm 21:59:29 OSTEP Team + pwhalen arriving Thursday 2pm at MCI. 21:59:35 I'll create a section for each of these so its easier to keep track of 22:00:06 No need to duplicate arrival info, there's already a FUDCon page for that. 22:00:38 * masta arriving late wednesday 22:00:38 oh, then I should add my information there 22:01:23 is anyone arriving late? on Friday for example? 22:02:19 #topic 4) FUDCon Planning - c) Fedora ARM team dinner 22:02:33 thursday evening, for me 22:03:01 it should be friday or saturday 22:03:23 does anyone have anything that may conflict on either of those days? 22:03:52 Well, there's FUDPub 22:03:55 * jonmasters_ will have a limited budget to assist with covering dinner for the core Fedora ARM team 22:04:02 fudpub will probably be saturday, no? 22:04:43 Looks like it. ianweller - can you confirm? 22:04:48 jonmasters_: i can help pitch in 22:04:49 suggest we schedule on the day that isn't fudpub 22:05:05 fudpub is (i'm like really sure) saturday 22:05:06 So Friday 22:05:17 friday is barcamp, saturday/sunday is hackfests 22:05:50 ianweller: Recommendations for location for ARM team dinner Friday? 22:05:52 [additional note for pbrobinson et al, dmarlin_ tested the 3.8 kernel on highbank in HSV just now and it didn't get get beyond "Starting kernel" - we've some work to do] 22:05:59 ctyler: hmm. 22:06:21 ianweller: there are a couple of nice restaurants in downtown Lawrence I liked the look of 22:06:27 free state brewery is always good but call ahead 22:06:41 ianweller: there's a place next to the homemade ice cream shop on the main strip that was good 22:06:51 * jonmasters_ is strugging with the name 22:07:02 oh next to sylas and maddy's? 22:07:06 yea 22:07:25 there's an Indian or Japanese place there. Actually, there were a few goodish looking places IIRC 22:07:33 what kind of food are we after? 22:07:46 italian! 22:07:49 local 22:07:49 it is also possible to do Kansas City, MI for dinner - it's not too far and we can drive 22:07:56 s/MI/MO/ :) 22:08:07 oh indeed 22:08:08 * ianweller thought there was a kansas city in michigan suddenly 22:08:10 im fine with anything 22:08:19 I was texting someone in MI earlier, sorry :) 22:08:33 45m drive either way. bleah. 22:08:46 bconoboy: well, it's not that bad, it's one highway 22:09:05 might be rush hour considerations. I'd suggest staying local. 22:09:09 I'm just saying Kansas City is actually very beautiful and so you might enjoy a little touristy dinner trip 22:09:11 Next to Sylas & Maddy's is ... Fatso's? 22:09:21 we don't really want to go to fudcon to run away from fudcon 22:09:32 bconoboy: +1 22:09:33 i just cant eat a huge stack of pancakes tho 22:09:34 well, fair enough 22:09:38 only a couple 22:09:45 there's plenty of good stuff in town. i'm just a bad guy to ask for recommendations :P 22:09:55 because without fail i just end up drawing a blank 22:09:56 ianweller: I'll ask Rikki for input 22:10:05 jonmasters_: that's a good idea. 22:10:22 ianweller: which day is fudpub? saturday? 22:10:27 #action jonmasters to sync with a friend in Lawrence on dinner options (with bias toward Italian options) 22:10:28 bconoboy: afaik yeah 22:10:38 :P 22:10:39 #action arm team dinner on friday 22:10:49 * ianweller is double checking that 22:11:04 ianweller: I'm staying at Rikki's place btw 22:11:43 i wonder if she's planning on coming to the conference at all, i don't see her on the prereg list 22:11:53 I intend to drag her over 22:11:55 :) 22:12:11 #topic 5) Your topic here! 22:12:16 soooo 22:12:20 jonmasters_: sounds good to me :) 22:12:25 where in OT now, so anything else for today? 22:12:28 koji 1.7.1 doesn't seem to work for me right now 22:12:36 fudpub is definitely a saturday thing 22:12:38 but koji 1.7.0 was working fine i believe 22:12:41 * ianweller disappears 22:13:08 f18/rawhide on raspberry pi? 22:13:08 And that's a wrap. Thanks all! 22:13:24 im trying to get f18 built but koji-1.7.1 is holding me up :/ 22:13:33 ctyler: :-) 22:13:34 until i fix it or downgrade it i can try 22:14:12 there is an f18-v5 we released i believe 22:14:20 #endmeeting