#title #ubuntu-meeting Meeting Meeting started by NCommander at 15:01:04 UTC. The full logs are available at http://ubottu.com/meetingology/logs/ubuntu-meeting/2011/ubuntu-meeting.2011-09-15-15.01.log.html . == Meeting summary == ''LINK:'' https://wiki.ubuntu.com/ARM/Meeting/2011/20110915 (NCommander, 15:01:39) *Action Item Review *ogra to find out whats wrong with the tracker *Standing Items *http://people.canonical.com/~platform/workitems/oneiric/ubuntu-arm.html ''LINK:'' http://people.canonical.com/~platform/workitems/oneiric/ubuntu-arm-ubuntu-11.10-beta-2.html (NCommander, 15:03:19) ''LINK:'' https://wiki.ubuntu.com/ARM/PBlueprintIdeas (GrueMaster, 15:07:52) ''LINK:'' https://wiki.ubuntu.com/ARM/PBlueprintIdeas#preview (NCommander, 15:07:53) *ARM Server Status (NCommander, Daviey) *Kernel Status (cooloney, ppisati) *ARM Porting/FTBFS status (NCommander, janimo) *ARM Image Status (ogra, NCommander) *QA Status (GrueMaster, mahmoh) *Linaro Updates (rsalveti) *AOB Meeting ended at 16:14:16 UTC. == Votes == == Action items == * (none) == People present (lines said) == * ogra_ (163) * GrueMaster (58) * NCommander (44) * janimo (39) * rsalveti (24) * ppisati (23) * rbasak (8) * mahmoh (7) * davidm_ (6) * meetingology (3) * ubottu (2) * Ursinha (2) * Daviey (1) == Full Log == 15:01:04 #startmeeting 15:01:04 Meeting started Thu Sep 15 15:01:04 2011 UTC. The chair is NCommander. Information about MeetBot at http://wiki.ubuntu.com/AlanBell/mootbot. 15:01:04 15:01:04 Available commands: #accept #accepted #action #agree #agreed #chair #commands #endmeeting #endvote #halp #help #idea #info #link #lurk #meetingname #meetingtopic #nick #progress #rejected #replay #restrictlogs #save #startmeeting #subtopic #topic #unchair #undo #unlurk #vote #voters #votesrequired 15:01:07 fnop 15:01:14 fput() 15:01:19 :) 15:01:20 * davidm_ waves 15:01:39 #link https://wiki.ubuntu.com/ARM/Meeting/2011/20110915 15:01:49 [topic] Action Item Review 15:01:58 [topic] ogra to find out whats wrong with the tracker 15:02:12 thats moot 15:02:19 we had the wrong url 15:02:34 ah, so a PEBKAC problem 15:02:48 * NCommander can't spell this mornming 15:02:51 heh 15:03:00 [topic] Standing Items 15:03:08 [topic] http://people.canonical.com/~platform/workitems/oneiric/ubuntu-arm.html 15:03:19 [link] http://people.canonical.com/~platform/workitems/oneiric/ubuntu-arm-ubuntu-11.10-beta-2.html 15:03:58 hello 15:04:25 * ogra_ has a question for one of his items ... 15:04:38 do we actually want to do the ext4 switch right now ? 15:05:02 it seriously seems a bit late to me for such a change 15:05:04 for preinstalls? 15:05:07 yes 15:05:16 live-build only recently got support for ext4 15:05:25 ogra_, would be good imo, does live-build have the patch already? (In Ubuntu I mean) 15:05:29 so we could theoretically still do the switch 15:05:32 Should have been done for A2/A3. Little late now. 15:05:37 but that also means lots of fallount i think 15:05:46 Best to stay with ext3 for now, and change it for A1 in P cycle 15:05:53 janimo, no idea, but we could pull it in 15:06:04 but i'm more in a tendency to postpone it 15:06:29 ogra_, not a very important thing, just a would be nice to align with ubuntu x86 and linaro 15:06:36 yeah 15:06:38 i agree 15:06:48 definately something for next cycle 15:06:52 but beta freeze is tomorrow and the images still have tons of bugs 15:06:55 with linaro's monthly schedule we will always have stale images featurewise for a few months 15:06:57 * NCommander adds it to the blueprint list 15:06:59 sure 15:07:28 NCommander, blueprint wiki link please? 15:07:35 Beta freeze is today. 15:07:51 So it is not a real freeze, just a beta one 15:07:52 [link]https://wiki.ubuntu.com/ARM/PBlueprintIdeas 15:07:53 https://wiki.ubuntu.com/ARM/PBlueprintIdeas#preview 15:08:15 drat, GrueMaster was faster 15:08:20 GrueMaster, well, past midnight for me ;) 15:08:31 NCommander, GrueMaster was cleaner :P 15:09:12 [topic] ARM Server Status (NCommander, Daviey) 15:10:30 server itself looks like its in good shape for A2, though we have some outstanding OMAP issues which I'll bring up during porting 15:10:54 anyone got anything else? 15:12:00 is mx5 related to server in any way? 15:12:17 no 15:12:24 we dont roll server for it 15:13:34 Guess there's nothing else 15:13:45 [topic] Kernel Status (cooloney, ppisati) 15:13:54 it's me, Paolo! :) 15:14:00 yay 15:14:07 there;s a new omap4 kernel 15:14:39 and this new kernel fixes some outstanding issues: framebuffer coloring, audio, and a couple of warnings on boot 15:14:53 there's a new ac100 kernel upload preparing to fix the USB speed issue and to enable DRM but nothing else 15:15:04 framebuffer coloring ? are you serious ?!? 15:15:08 unfortunately the new kernel breaks the pvr driver, but that should be fixed in an upcoming release 15:15:12 yep 15:15:12 :) 15:15:18 * ogra_ got used to the blue after all these releases 15:15:22 no more "smurf-colored" console 15:15:23 :) 15:15:26 the USB speed issue is fixed already in the omap4 kernel? 15:15:30 oh yes 15:15:30 yes 15:15:31 sorry 15:15:37 i forgot about the important stuff 15:15:51 I thought purple in d-i was intentional 15:15:52 yes, Ming's ehci fix is in there too 15:16:03 so, good usb i/o too 15:16:17 NCommander, but blue text in black tty console wasnt ;) 15:16:30 now we "only" need to fix pm and pvr driver and that will be the best omap release so far :) 15:16:43 oh yes, and kexec too... 15:16:45 wohoo 15:17:09 on the ac100 front, janimo takles over the ac100 kernel now 15:17:26 so we will finally get a git tree i guess :) 15:17:34 * ppisati didn't mention the usual round of kernel updates for all the other arm flavours since we are getting one of this every week 15:17:43 cool 15:17:51 yay git 15:17:58 pfft git 15:18:01 lol 15:18:13 * Ursinha grabs the popcorn and watches 15:18:19 lol 15:18:30 and it seems cooloney is back 15:18:48 oh, nice, will he come back and attend then ? 15:18:52 yay git 15:18:53 2:1 15:19:01 ah, dunno 15:19:08 rbasak: Around? 15:19:19 perhaps he will move to West Europe/US? :) 15:20:09 Daviey: yes! 15:20:13 anytihng else or should move on? 15:20:22 move 15:20:31 go 15:20:42 [topic] ARM Porting/FTBFS status (NCommander, janimo) 15:20:49 horrid ! 15:20:54 wrt universe 15:21:03 given back a few packages 15:21:05 main looks okayish 15:21:17 i went through all ftbfs logs the last days 15:21:20 nothing actually changed/fixed but that 15:21:35 70-80% of the arm ftbfs are GLES stuff 15:21:45 On the OMAP3 side, netboot remains broken, fix there should be forthcoming today 15:21:46 some libavcodec 15:21:47 we probably get to fix pgsql unless upstream is on it 15:22:03 and a bunch of qreal 15:22:26 ogasawara, 70-80? I doubt it is that much. I would have thought it is around 30% 15:22:29 needs to be counted :) 15:22:32 rsalveti, do you have any std reciepe for GLES issues ? 15:22:50 * rsalveti reading 15:22:55 Guys, check your name tab completion. 15:22:56 the ftbfs wikipage could need some code examples etc 15:22:56 ogra_, I think those apps need porting to GLES or have that code disabled. 15:23:10 ogra_: not for now, planned to discuss at UDS 15:23:16 it is very app specific 15:23:20 because the apps need porting 15:23:24 janimo, well, if its just: change dep, switch configure options and build ... 15:23:38 some apps have gles support 15:23:46 but haent got it enabled 15:23:51 on OMAP4, I've been grinding away at updating partman-uboot/partman-reciepes for the last week and some before I went for vacation 15:23:54 * ogra_ resarched some of them 15:24:13 Its 95% working but still has some odds and ends. I've talked to skaet about it, and I'm expecting it to land tonight/tomorrow 15:25:24 ogra_, the Qt ones cannot work as they use GLES and OPENGL in the same time 15:25:28 and those are quite a few 15:25:34 right 15:25:52 i mean the ones that are ported upstream but where the build behaves like on x86 15:26:11 don't believe we have much of those still 15:26:18 not many, but a few 15:26:24 most of the issues are from the qt gl x gles issue 15:26:59 stellarium is a prominent one 15:27:06 and there are a few others 15:27:25 anyway, i dont want to slow the meeting 15:27:37 NCommander, move :) 15:28:04 [topic] ARM Image Status (ogra, NCommander) 15:28:27 so so ... 15:28:48 we had some issues with the build system that hopefully shoudl be resolved 15:29:30 lamont, replaced something on annonaceae since then omap and mx5 builds seem to hang 15:29:48 i pinged him but he doesnt seem to be around, i'll hunt down someone else from IS later 15:30:06 ac100 images build and install fine, mx5 should be building fine too now 15:30:23 ogra_, so the lack of today's images is due to above breakage? 15:30:32 oh and archive skew costed us three days of images this week 15:30:33 indeed ac100 is working fine 15:30:47 nice, less than it costs on average :) 15:30:56 * NCommander whacks janimo 15:31:02 janimo, yes, annonaceae doesnt build anything apparently, i suspect an ssh key was replaced or so 15:31:46 NCommander, anything to add for the server or netinst images 15:31:50 ? 15:32:08 just the previous methoned partman-uboot hacking for OMAP3/4, and the f-k fix 15:33:08 move then :) 15:33:28 [topic] QA Status (GrueMaster, mahmoh) 15:33:54 All kernel SRU testing for systems on life support completed this week. 15:33:54 Most of the testing for server is no finished, at least for a once over. 15:34:46 My only beagleXM has an odd issue with the latest x-loader/u-boot where it will no longer connect to the network. Driver loads, but it will not show a link. 15:34:54 Using the Natty versions works fine. 15:35:09 GrueMaster: were you able to test with linaro's kernel? 15:35:33 my xm rev a1 works ok with O u-boot 15:35:42 not yet. Had to reimage (I was running a cryptfs and linaro doesn't support that). 15:36:04 Will test linaro kernel today. 15:36:20 MMlosh in #ubuntu-arm reported the same NIC issue 15:36:31 Also plan on doing some deep desktop image testing for the rest of the week, as I have done little on that this cycle. 15:36:31 but he is on natty and just using the oneiric kernel 15:36:51 interesting. 15:36:55 so i'm not sure it isnt userspace 15:37:07 but indeed intresting that you see the same 15:38:08 mahmoh: anything to add? 15:38:26 GrueMaster: nothing except that you're doing a bang up job 15:38:33 ++ 15:38:48 and he is also our best supporter in #ubuntu-arm 15:38:52 [topic] Linaro Updates (rsalveti) 15:38:58 * rsalveti waves 15:39:27 well, not much that touches oneiric 15:39:33 first version of gcc-linaro packages 15:39:39 to be available at the toolchain backports PPA 15:39:51 to help testing toolchain issues and validating the gcc-linaro later on 15:40:08 on-going debugging on LTTng 2.x 15:40:15 first time we're trying to make it work with ARM 15:40:27 good thing is that we don't have a lot of kernel dependencies as before 15:40:35 hope to make it work properly this time 15:41:09 Is there a plan to integrate x-loader with u-boot for omap next cycle? Also, PXE would be nice, since the code is there. 15:41:09 looking to try to get bug 669641 fixed at the linaro-kernels atm 15:41:10 Launchpad bug 669641 in linux (Ubuntu) "systemtap fails to discover installed debug modules" [Undecided,Triaged] https://launchpad.net/bugs/669641 15:41:25 then push it to ubuntu-kernel once the fix is available and working 15:41:30 nice 15:41:39 GrueMaster: yup, in theory SPL already works for OMAP 15:41:48 we'll validate it for this cycle, so for P it should work the same way 15:41:59 bye bye xloader 15:42:04 yeah 15:42:11 well, not completely 15:42:16 yet 15:42:43 but at least for next cycle for sure :-) 15:42:53 anything else? 15:43:01 and on-going multi-arch fixes in place to get firefox cross buildable 15:43:02 well you still keep MLO 15:43:23 sure, but the main issue was how x-loader was maintained 15:43:30 that's the next step 15:43:32 indeed 15:43:33 ogra_, yes, but no more extra packages that we need to keep remembering the names of 15:43:41 NCommander, nothing from me 15:43:44 that's most from my side 15:44:09 thanks rsalveti 15:44:18 [topic] AOB 15:44:27 erm 15:44:48 what about blueprints ?? werent we supposed to have a topic for that today ? 15:44:51 erm? 15:45:04 * ogra_ thought davidm wanted to discuss them today 15:45:09 Yes, I did want to see a review of bp 15:45:23 I'm on a call that I can 't break from 15:45:41 davidm, lets do it on the call then 15:45:55 for the community: https://wiki.ubuntu.com/ARM/PBlueprintIdeas oin case you have ideas to add for us :) 15:46:05 But I would like to see the team review the BP page https://wiki.ubuntu.com/ARM/PBlueprintIdeas 15:46:21 lets go over them then 15:46:28 The page is scheduled to lock soonish 15:46:50 say 19th at the top schedule :) 15:46:55 *says 15:47:18 ah, no, even 3rd 15:47:25 Actually, I think it is the 3rd that it locks. 15:47:29 yeah 15:47:41 i slipped a row 15:47:52 forst one is "Implement an ARM Server Kernel for Server Hardware" 15:48:07 At any rate. Shall I list them and we can beat^h^H^Hdiscuss them? 15:48:10 Is that a kernel flavour, like server and virtual? 15:48:31 Server kernel makes sense once we have a server platform. 15:49:15 will this have to be per platform? 15:49:18 i guess that actually depends on having server hardware :) 15:49:18 GrueMaster, oh, feel free 15:49:18 * ogra_ steps back 15:49:20 rbasak, yes, the -server kernel uses a few different config options 15:49:35 mahmoh, well, it hads to be for the "server platform" 15:49:55 it doesnt make sense to put a fulltime person on a server kernel without having server HW 15:50:06 and arm should have its own configs on top of that 15:50:10 and i guess it will be a good part of a fulltime job 15:50:37 (split across kernel and QA) 15:51:02 Arm Server kernel build configuration and tuning -- Should combine with previous. 15:51:09 i think we all agree that we want such a kernel but we need HW first 15:51:22 else it is a waste of efforts 15:51:32 GrueMaster, ++ 15:51:49 Unified OMAP3/OMAP4 install images 15:51:50 though we should probably have a general kernel configuration session 15:52:33 (covering the config handling for arm in general) 15:52:48 How would this work? Same kernel/u-boot? Is it possible? 15:52:51 unified images can only work if we have unified MLO and u-boot 15:53:00 And kernel. 15:53:06 kernel should already be possible i think 15:53:23 but indeed not if we go on with the TI patches which we have to 15:53:29 * All ARM headless images are based on Ubuntu Core, 15:53:38 * ogra_ just nods 15:53:39 GrueMaster, kernel is unified in linaro 15:53:41 move :P 15:53:51 uboot to be explored, part of the purpose of the BP 15:54:03 oops, missed the wrap. 15:54:05 we dont have headless atm ... if it comes back it will be based on core ... done 15:54:12 All ARM headless images are based on Ubuntu Core, 15:54:12 except for specific ARM Server Images 15:54:25 not really a spec 15:54:38 rather a "this is what we will use if it comes back" 15:54:38 * Explore Codezero Embedded Hypervisor Virtulazation on A9 http://www.l4dev.org/ 15:54:47 ok 15:55:08 This could be fun. 15:55:13 yeah 15:55:29 * Server Performance Testing incl Storage 15:55:48 well, thats just a carry over of the server QA from O 15:55:51 or not ? 15:56:00 or an extention to it 15:56:07 I think storage should be broken out, otherwise this will get big 15:56:08 Need more detail. mahmoh has already been running the phoronix test suite to shake out issues there. 15:56:46 Storage will get tested properly once we have hw that has better support. 15:56:55 * ogra_ is sad that there are so many server specs but not many tech improvements or community/desktop specs ... 15:57:12 * ARM Hard float 15:57:21 is that a spec ? 15:57:24 We're about to runout of time sadly 15:57:33 It's because we have already rocked the desktop. 15:57:34 we surely need discussion about it 15:57:48 move to ubuntu-arm ? 15:57:52 GrueMaster, well, enough ? is the question 15:58:03 well, is there a meeting after us ? 15:58:16 but yeah, else move to -arm and finish that 15:58:17 I'm not sure 15:58:27 * Orchestra/Ensemble on ARM 15:58:42 * ogra_ wonders whats arm specific there 15:59:12 deployment issues apparently. 15:59:14 * Orchestra/Juju on ARM 15:59:15 but then i dont know enough about it ... though i thought its a management tool 15:59:19 noone elseis here (and actually, the frdge event for the meeting still says Mobile and Tuesday) 15:59:27 lol 15:59:32 * Linaro ARM Boot 15:59:40 whats that ? 15:59:58 BTHOOM 16:00:00 there are a bunch of boot related things in our build system i would like to change 16:00:08 (beats the hell out of me) 16:00:27 boot speed maybe? 16:00:30 so having some info what "linaro boot" means would be nice 16:00:33 * KVM on ARM 16:00:35 * GrueMaster cringes 16:00:39 hehe 16:00:44 you asked for it :P 16:01:07 we could also add "libvirt on arm" :) 16:01:18 No, I didn't. I asked for tests to be written for bare metal, then adapted for kvm, not vice versa. 16:01:22 we should add libvirt 16:01:34 argh, that was a joke ! 16:01:43 heh 16:01:49 * ogra_ looks for duct tape for his nasty mouth 16:01:50 * Server System management on ARM - IPMI 16:01:50 It needs to happen eventually! 16:02:05 how is that different to the juju spec ?? 16:02:24 rbasak: So does a solution for bug 1 16:02:27 Launchpad bug 1 in Ubuntu "Microsoft has a majority market share" [Critical,In progress] https://launchpad.net/bugs/1 16:02:28 does it make sense to have two of them ? 16:02:33 :-) 16:02:37 i think we could merge them in one 16:02:44 same for kvm/libvirt 16:03:26 * QA Develop automation tests that can run without kvm support (i.e. on raw hardware). 16:03:44 har! 16:03:47 hehe, attacking the kvm spec :) 16:04:09 spec to spec combat! 16:04:30 There is a plan to help with that I think 16:04:38 I think this is beneficial, because I am always finding bugs on raw hw that don't manifest on kvm. 16:04:45 Get nova to be able to launch metal instances from a pool of available metal 16:04:46 yeah 16:04:59 (maybe via LXC) 16:05:00 sounds trivial 16:05:26 * Revisit Server QA tests from Oneiric, enhance and update as needed. 16:05:28 given that we already implemented that :) 16:05:35 (as the panda cluster) 16:05:50 does that use nova? 16:06:00 never heard of nova 16:06:01 A lot of the tests this round were random (and some have no tests in any arch). 16:06:35 well, iirc in budapest is was said that this cycle was to determine a better set of tests 16:06:50 so thats an expected follwoup i guess 16:07:36 * Compare kernel configs with x86/amd64 configs for for module differences (NFS, VFAT, etc) 16:07:51 shoudl become a general kernel config spec 16:08:01 for arm indeed :) 16:08:17 I ran into a lot of qa kernel SRU tests that fail simply because some modules are compiled =Y as opposed to =m. 16:08:43 And I also had to postpone some tests because of missing modules. 16:08:49 yeah 16:09:03 and we have a good bunch of differences to x86 as well 16:09:20 so we should do a general review 16:09:42 * One consumer device per cycle programme 16:09:48 yay 16:10:00 so my idea is that we enable one enduser device per cycle 16:10:12 preferably one that is close to images we already have indeed 16:10:23 and that the community can chose which device that is 16:10:28 GrueMaster, would be good to have a list of features failing unless modules 16:10:32 I disagree on this, only because we got bogged down on 1 and played hot potato with the other. 16:10:40 I was thinking of trying to build in as many as possible for bootup speed 16:10:48 GrueMaster, bogged down ? 16:10:57 but if it some do not work it'd be good to know ahead 16:11:05 we have ac100 images now ... and we could have nook ones if you had had the time 16:11:08 How long have you been working on ac100? 16:11:25 the plan is to deeply involve the community here 16:11:26 we could have 5 images per cycle if we had better build tools honestly 16:11:30 ac100 isnt comparable 16:11:42 i had to invent a totally new image type for it 16:11:46 the time spent is not as much coding but debugging the innards of the build system 16:11:46 Problem with nook images is lack of touch screen keyboard in ubiquity. 16:11:58 eaily solved :) 16:12:05 *easily 16:12:09 you didnt tell me 16:12:20 the idea is a good one, maybe choose a family of similar products and then a target to work from 16:12:26 I was just saying once we get the tools shacken out, we shouldn't limit it to one. 16:12:33 ogra_, I doubt the community has better chances of getting something up with cdimage/debian-cd as they are used now 16:12:36 anyway, we can discuss that later, thats my spec proposal and the masterplan is actually to grow community 16:12:44 +1 16:12:45 but we need the community for upstream work (kernel etc) and testing 16:12:48 Sounds good. 16:12:56 janimo, thats the part we should be offering 16:13:08 debian-cd/cdimage stuff 16:13:23 as they are now they are... juju 16:13:26 the community still needs to supply the external bits 16:13:55 anyway, lets do details later and let NCommander close the meeting :) 16:14:07 closing in 3 16:14:09 2 16:14:13 1 16:14:16 #endmeeting Generated by MeetBot 0.1.5 (http://wiki.ubuntu.com/AlanBell/mootbot)