20:00:28 <pwhalen> #startmeeting Fedora ARM weekly status meeting
20:00:28 <zodbot> Meeting started Wed Sep  4 20:00:28 2013 UTC.  The chair is pwhalen. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:00:28 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
20:00:28 <pwhalen> #chair pwhalen jonmasters bconoboy ctyler pbrobinson dgilmore dmarlin masta handsome_pirate msalter ahs3 agreene jcapik ddd
20:00:28 <zodbot> Current chairs: agreene ahs3 bconoboy ctyler ddd dgilmore dmarlin handsome_pirate jcapik jonmasters masta msalter pbrobinson pwhalen
20:00:32 <ahs3> greetings, fellow earthlings.
20:00:35 <pwhalen> .fas pwhalen
20:00:36 <zodbot> pwhalen: pwhalen 'Paul Whalen' <pwhalen@redhat.com>
20:00:42 <masta> m000000000
20:00:43 <bconoboy> .fas blc@
20:00:44 <zodbot> bconoboy: blc '' <blc@redhat.com>
20:00:46 <pwhalen> afternoon everyone :)
20:00:53 <ahs3> .fas ahs3
20:00:53 <zodbot> ahs3: ahs3 'Al Stone' <ahs3@redhat.com>
20:00:58 <msalter> .fas msalter
20:00:58 <zodbot> msalter: msalter 'Mark Salter' <msalter@redhat.com>
20:02:39 * handsome_pirate waves
20:02:48 <masta> few minutes for folks to fall into rank/file
20:03:20 <pwhalen> alright, lets get started
20:03:23 <oatley> .fas oatley
20:03:23 <pwhalen> #topic 1) Problem packages
20:03:23 <zodbot> oatley: oatley 'Andrew Oatley-Willis' <andrew.oatley-willis@senecacollege.ca>
20:04:16 <pwhalen> anyone aware of any build problems?
20:04:41 <hrw> pwhalen: on armv7hf?
20:05:11 <bconoboy> pwhalen: this topic might not make sense as much any longer now that armv7hl is part of PA
20:05:33 * jsmith is late to the party
20:05:33 * bconoboy thinking out loud
20:05:34 <hrw> pwhalen: Remi Collet wrote me that php has some failed tests on arm
20:05:35 <pwhalen> bconoboy, agreed
20:05:56 <jcapik> .fas jcapik
20:05:56 <zodbot> jcapik: jcapik 'Jaromír Cápík' <jcapik@redhat.com>
20:06:09 <bconoboy> pwhalen: we could either open it up to aarch64 or change the focus to packages that build but are broken :-)
20:06:45 <pwhalen> we'll move on then for now, re-evaluate for next week, we have aarch64 on the agenda
20:06:48 <pwhalen> #topic 2) Kernel Status Update
20:06:57 <pwhalen> the fun one!
20:06:58 <hrw> o yes
20:07:24 <jsmith> It's been a busy couple of weeks in kernel land, it seems :-)
20:07:31 <pwhalen> #info kylem has managaed fix our PCIe issues on trimslice, official build pending
20:07:41 <masta> so I hear we got tegra2 pci solved
20:07:53 <bconoboy> sounded like the next kernel build would also include bbb fixes
20:07:58 <jsmith> I know pbrobinson has been working hard on the Beagles too
20:08:15 <pwhalen> another win
20:08:27 <masta> bbb boots, but possibly problems with usb hub
20:09:03 <pwhalen> #info BeagleBone Black boots with possible USB issues
20:09:08 <pwhalen> has anyone tested the white?
20:09:20 <masta> Peter test both white and black
20:09:25 <pwhalen> cool
20:09:25 <masta> they are booting
20:09:27 <dgilmore> pwhalen: kernel is not built yet so its not testable
20:10:19 <pwhalen> kylem, anything you wanted to discuss?
20:10:26 <bconoboy> #undo
20:10:26 <zodbot> Removing item from minutes: <MeetBot.items.Info object at 0x27b91c10>
20:10:31 <kylem> no, sorry, was looking at something else
20:10:37 <bconoboy> #info Next kernel build will include fixes for trimslice and beaglebone black
20:10:49 <pwhalen> bconoboy, thanks
20:10:54 <hrw> is there up-to-date list of will-be-supported devices?
20:11:15 <masta> I'm looking forward to retire the BBB remix once the kernel is happy
20:11:17 <kylem> highbank. everything else is best effort. ;-)
20:11:45 <bconoboy> Current list (for alpha) appears to be highbank, qemu, wandboard (?), trimslice (pending), maybe beaglebone xm, maybe beaglbone black
20:11:59 <hrw> bconoboy: s/beaglebone xm/beagleboard xm/
20:12:07 <bconoboy> uh, yeah.
20:12:12 <bconoboy> beagle.*xm
20:12:17 <hrw> ;D
20:12:33 <hrw> I do wonder do my beagleboards c3/b7 still work
20:13:20 <hrw> bconoboy: qemu as 'virtual express with A9' or?
20:13:30 <pwhalen> #topic 3a) Aarch64 - Status Update, problem packages
20:13:33 <bconoboy> hrw: correct
20:13:55 <bconoboy> Current status: 12045 packages built
20:14:00 <hrw> pwhalen: acpid should be dropped from list
20:14:02 <bconoboy> Queue size: 0
20:14:30 <bconoboy> I'd like to take a moment to introduce hrw, also known as Marcin Juszkiewicz, who has joined Red Hat to work on aarch64 packages
20:14:38 <hrw> how to get to build logs? as daily summary only gives list of failed packages
20:14:47 * hrw \o/
20:14:52 <bconoboy> hrw: Let's discuss after the meeting, I'll get you setup
20:14:57 <hrw> bconoboy: mkey
20:15:00 <bconoboy> #info 12045 packages built
20:15:25 <bconoboy> #info hrw, aka Marcin Juszkiewicz, has joined Red Hat to get the last couple thousand built :-)
20:15:26 <hrw> bconoboy: I remember day one of OE aarch64 bootstrap... ~1000 packages built
20:15:43 <bconoboy> hrw: it's the last 10% that take 90% of the time...
20:15:50 <hrw> bconoboy: as usual
20:16:01 <hrw> bconoboy: and built != working ;(
20:16:10 <bconoboy> hrw: True, but it's progress...
20:16:17 <hrw> .fas juszkiewicz
20:16:18 <zodbot> hrw: hrw '' <marcin-rhbugz@juszkiewicz.com.pl>
20:16:30 <bconoboy> Anyway, that's my status for the week
20:16:35 <bconoboy> Maybe next week hrw will provide :-)
20:16:43 <pwhalen> thanks bconoboy
20:17:03 <hrw> bconoboy: hope for it
20:17:18 <hrw> bconoboy: s/:/: I/
20:17:35 <pwhalen> #topic 3b) Aarch64 - Koji status
20:18:03 <bconoboy> dgilmore/nirik: Any update on getting the aarch64 koji server going?
20:18:14 * masta wants to learn how to pronounce juszkiewicz
20:18:17 <bconoboy> It's troubling that we're near F20 alpha and have not yet begun F20 builds on aarch64.
20:18:39 <ddutile> masta: just like it's spelled !
20:18:43 <hrw> masta: if you speak English natively: s/sz/sh s/cz/ch and if may sound acceptable
20:19:15 <masta> ok, just want to make sure not to say it wrongly
20:19:24 <bconoboy> I'd like to suggest we move on to stage5, which is F20, using the arm-rebuild.sh script if nothing else is going to happen on koji for the next few weeks.
20:19:25 <masta> acceptible is good
20:19:29 * ddutile asks dzickus for how to pronounce Polish names...
20:19:43 <hrw> masta: find some Polish/Russian/Czech/Slovak guy and you may get example
20:20:04 <jcapik> it's juškievič in czech :)
20:20:16 <pwhalen> does anyone disagree with moving to stage5?
20:20:23 <msalter> that helps :)
20:20:34 <hrw> bconoboy: "queue size: 0" means "there is nothng in build queue at the moment" right? so switch should just work
20:20:53 <handsome_pirate> pwhalen:  No arguments here
20:20:55 <hrw> jcapik: your š is softer than our sz
20:20:58 <bconoboy> hrw: Yeah, nothing in the queue. The backstory is that we want to move to the standard fedora build system infrastructure, which is koji
20:21:23 <handsome_pirate> pwhalen:  The problem is there isn't enough Infra folks to go around
20:21:24 <bconoboy> hrw: For stage4 we're just using a hacky shell script to distribute builds, it'd be good to leave it behind
20:21:37 <bconoboy> dgilmore/nirik: ping?
20:21:53 <dgilmore> bconoboy: yes
20:21:55 <hrw> bconoboy: and koji knows about 'package xyz is not for this arch so skip' as well
20:22:14 <dgilmore> bconoboy: not had a chance to look,and really dont know when i will
20:22:30 <bconoboy> dgilmore: eta to setting up aarch64 koji server?  we talking weeks or months?  I think, if more than 2 weeks, it's time to move to f20 builds.
20:22:31 <dgilmore> hrw: thats embeded in the spec file
20:22:43 <dgilmore> bconoboy: no idea
20:22:44 <hrw> dgilmore: and stage4 ignores that
20:22:53 <dgilmore> hrw: it does
20:23:11 <bconoboy> hrw: Yes, koji groks the spec file internals, whereas the queueing script is quite dumb and does not.
20:23:38 <dgilmore> the queueing script was intentinally very simple and light weight
20:23:49 <hrw> bconoboy: I would go for koji then. you guys bootstrapped enough to get builds working
20:23:51 <bconoboy> dgilmore: Okay, so let's say I create a stage5, and pwhalen updates the builders to use it...
20:23:59 <msalter> we also disable the %check in stage4 to speed up builds on the models, so well probably find some test failures (which is a good thing)
20:24:15 <bconoboy> dgilmore: Anything we need to do or not do to make sure importing the resulting stage4/stage5 hybrid into koji isn't extra painful?
20:25:50 <bconoboy> dgilmore?
20:26:19 <dgilmore> bconoboy: i really dont understand you
20:26:29 <bconoboy> dgilmore: ah. let me rephrase.
20:26:38 * ddutile too... :-(
20:27:09 <bconoboy> dgilmore: Some day, we want to import the result of all this bootstrap into koji, right?  Right now we have a bunch of rpms in stage4 that are either imported from f19 (noarch), or rebuilt.  They all have an fc19 tag on them
20:27:46 <bconoboy> dgilmore: Hypothetically, if we start building up f20 in stage5, are there any special measures to be taken to ensure that the result can be easily imported into koji, once the server is available?
20:28:08 <dgilmore> bconoboy: no, because we dont import any of it into koji
20:28:21 <bconoboy> ?
20:28:35 <dgilmore> bconoboy: we add the existing builds as an external repo and rebuild them all in koji
20:28:56 <bconoboy> dgilmore: Ugh, so if we rebuild 90% of F20 in stage5 we have to rebuild that 90% all over again?
20:29:27 <dgilmore> bconoboy: yes
20:29:36 <bconoboy> that bites
20:29:38 <dgilmore> thats how things are designed
20:29:51 <dmarlin> so we need koji real soon now
20:29:53 <bconoboy> can somebody else do the koji setup then?
20:29:55 <dgilmore> bconoboy: it doesnt bite, its how it was designed to be done
20:29:59 <hrw> dgilmore: bootstrapping f20 from f19 and then rebuilding f20 from f20 bootstrap?
20:30:24 <dgilmore> bconoboy: nirik can
20:30:30 <bconoboy> dgilmore: Rebuilding F20 twice, on the foundation model, that bites.
20:30:33 <dgilmore> bconoboy: he has about as much time as I do
20:30:50 <bconoboy> dgilmore: Anybody else?  Is it just an access issue?
20:30:52 <jcapik> bconoboy: I wanted to help with the koji setup
20:31:03 <bconoboy> Both pwhalen and dmarlin have setup koji before, for instance
20:31:04 <dgilmore> bconoboy: the plan has always been build it once outside then rebuild it all again inside koji
20:31:06 <jcapik> bconoboy: but it still needs dgilmore and/or nirik involved
20:31:07 * nirik looks up. was busy with something else.
20:31:10 <dgilmore> you shouldnt be surprissed
20:31:13 * masta would like to help with koji too
20:31:15 <nirik> I can try and get this back on my list again.
20:31:28 <jcapik> at least the builders can be installed by us
20:31:30 <hrw> bconoboy: better on foundation one than on full one - you do not have to worry about emulated cpus stealing realcpu time just for core switching
20:31:44 <jcapik> things like certificates and tags can be made by dgilmore/nirik
20:31:54 <jcapik> i guess they have scripts for that
20:31:59 <nirik> I've never set it up from scratch... but I can try if there are people I can bug who have. ;)
20:32:02 <jcapik> shouldn't take so long
20:32:13 <dgilmore> nirik: its really not hard
20:32:14 <jcapik> I have some scripts from dhorak
20:32:29 <nirik> it should be setup in ansible. ;)
20:32:32 <bconoboy> Surely between all the volunteers here we can find a solution
20:32:43 <dgilmore> nirik: need to import the db schema and run a couple of commands to grant the initial admin user
20:32:44 <jcapik> they need to be modified to work with the latest koji, but .... hopefuly that's not gonna last too long
20:32:51 <bconoboy> I'd rather not waste a few hundred thousand cpu hours to rebuild twice though
20:33:04 <dgilmore> the trickiest bit will be setting up ansible
20:33:12 <dgilmore> bconoboy: stop
20:33:18 <nirik> bconoboy: let me try... not sure this week, but we are in freeze that gives me a bit more time often.
20:33:23 <dgilmore> bconoboy: thats the way that it has to be done period
20:33:39 <dgilmore> nirik: give you more and me less
20:33:45 <nirik> yep.
20:33:56 <bconoboy> dgilmore: if we build f20 in koji once why would we do it twice?  it'll be F21 time by the time it's done.
20:34:14 <dgilmore> bconoboy: your noty making sense now
20:34:28 <dgilmore> bconoboy: you're not making sense now
20:34:31 <bconoboy> dgilmore: :-)
20:35:17 <dgilmore> we need to rebuild the world again to make sure we get deps right
20:35:23 <nirik> really, let me try and get time to at least do initial setup. I'll try this week...
20:35:32 <bconoboy> nirik: Since we're in freeze and you have a little time, please please spend a bit of it on this.  We're losing ground every day- circular dependencies are reasserting themselves.  It really needs to be going or much of our work is wasted.
20:35:35 <dmarlin> nirik: +1
20:35:55 <nirik> bconoboy: freeze usually means I have more time, because since we are frozen I can't touch a bunch of our machines.
20:36:15 <bconoboy> :-)
20:36:26 <bconoboy> nirik: Okay, thanks.  I'll only pester you daily.
20:36:36 <nirik> I'm going to bug people who have set it up before tho most likely. :)
20:36:50 <nirik> fair enough. ping me in #fedora-arm and we can status there.
20:37:07 <bconoboy> nirik: Okay
20:37:22 <bconoboy> #agreed Nirik will start aarch64 koji setup this week as time allows
20:37:25 <nirik> also, if I can do initial perhaps I can hand off to someone else to do the rest.
20:37:43 <masta> =)
20:38:03 <bconoboy> pwhalen: next!
20:38:06 <pwhalen> nirik, cool, I can try and help. Thanks!
20:38:20 <pwhalen> #topic 5) FESCo Meeting summary - ARM to PA
20:38:28 <kylem> nirik: me three... i like knowing where the bodies are buried.
20:39:15 <pwhalen> there were some concerns from QA that we lacked enough working hw at the Alpha freeze
20:39:16 <nirik> cool.
20:40:10 <pwhalen> trying to find a link to the ticket if anyone has it handy
20:40:11 <handsome_pirate> So, the decision of FESCo is to let ARM and QA negoitiate blocking h/w
20:40:48 <pwhalen> as it is currently
20:40:52 <pwhalen> http://fedoraproject.org/wiki/Fedora_20_Alpha_Release_Criteria#Release-blocking_images_must_boot
20:40:56 <handsome_pirate> Tim is of the stance (and I agree) that highbank + 1 other platform (besides vexpress) working is acceptable
20:41:20 <handsome_pirate> 1 other *common* platform
20:41:34 <masta> cool!
20:41:38 <pwhalen> which links to these being supported in f20 - Trimslice , Highbank , Panda, Wand and Beagleboard
20:41:42 <bconoboy> Sounds like we should be there after the next kernel build
20:42:06 <bconoboy> I suggest we drop Pandaboard and Beagleboard from the blocking list.
20:42:18 <handsome_pirate> +1
20:42:20 <pwhalen> +1
20:42:23 <dmarlin> +1
20:42:28 <masta> +1
20:42:29 <masta> =(
20:42:33 <bconoboy> ... TI doesn't support it.  If it happens to work in 3.11, it's more luck than anything- probably be dead again in 2.12.
20:42:51 <handsome_pirate> proposed #agreed Drop Pandaboard and Beagleboard from blocking list
20:42:59 <handsome_pirate> Any acks?
20:43:06 <masta> uh...
20:43:14 <hrw> +1
20:43:17 <pwhalen> sure, this is sorta different for the arm folks
20:43:19 <pwhalen> ack
20:43:47 <handsome_pirate> One more?
20:44:01 <bconoboy> +1
20:44:05 <pwhalen> we've got buy in
20:44:08 <handsome_pirate> #agreed Drop Pandaboard and Beagleboard from blocking list
20:44:41 <pwhalen> anything else mentioned I missed?
20:45:02 <handsome_pirate> pwhalen:  That's the gist of the decision
20:45:16 <pwhalen> #topic 6) Open Floor
20:45:20 <kylem> bconoboy, i'm going to try the revert DT patch as an option for F-20
20:45:25 <bconoboy> pwhalen: would you make the update to the wiki?
20:45:27 <kylem> i have that built somewhere, just need to dig it up
20:45:33 <pwhalen> #info Fedora 20 Alpha TC3 available
20:45:40 <bconoboy> kylem: cool.  I'm just concerned it'll bite us with 3.12.
20:45:49 <pwhalen> #link https://dl.fedoraproject.org/pub/alt/stage/20-Alpha-TC3/Images/armhfp/
20:46:00 <bconoboy> Open floor question: What's the status of chromebook uboot with HYP support? Any success stories?
20:46:11 <pwhalen> please test and add your findings to the wiki
20:46:16 <handsome_pirate> bconoboy:  It's booting an upstream kernel
20:46:26 <pwhalen> bconoboy, sure
20:46:28 <handsome_pirate> bconoboy:  Building a virt-enabled kernel right now
20:46:30 <dgilmore> bconoboy: its irrelevant
20:46:36 <masta> I may turn panda into a remix if for some reason it continues to not work
20:46:38 <bconoboy> dgilmore: what?
20:47:05 <dgilmore> 20:45 < bconoboy> Open floor question: What's the status of chromebook uboot with HYP support? Any success stories?
20:47:24 <masta> hey
20:47:24 <bconoboy> dgilmore: er, how is that irrelevant?
20:47:30 <dgilmore> we really can not support the chromebook
20:47:47 <masta> so the linaro patches work for u-boot, but we have to add DMA patches to kernel, nto ack'ed upstream
20:48:10 <handsome_pirate> bconoboy:  There is a bug in the kernel such that an lpae kernel does not boot on exynos, trying a patch now to see if it works
20:48:19 <bconoboy> masta/handsome_pirate: Okay, so uboot with hyp = yes, working today?
20:48:23 <masta> and yes, as dgilmore said we cannto support the chromebook really... too hard for now with the way it boots
20:48:23 <handsome_pirate> bconoboy:  With lpae disabled, upstream kernel boots
20:48:37 <handsome_pirate> bconoboy:  With hyp uboot
20:48:39 <kylem> yo
20:48:50 <kylem> bconoboy, meh, it shouldn't be a problem
20:48:53 <masta> but I'm already starting to retool my system to make new f20 remix images
20:49:04 <hrw> handsome_pirate: where I can get chromebook uboot with hyp?
20:49:29 <masta> hrw: it's available in 2013.10-rc2
20:49:38 * handsome_pirate has to take off
20:49:44 <masta> I believe... might have to check that
20:49:47 <hrw> masta: yumi
20:49:48 <handsome_pirate> I'll be on again in a bit
20:49:48 <bconoboy> okay, thanks for the update.  that's it for me.
20:50:10 <bconoboy> #info linaro uboot 2013.10-rc2 has a chromebook uboot with hyp support
20:50:19 <dgilmore> #undo
20:50:19 <zodbot> Removing item from minutes: <MeetBot.items.Info object at 0x283456d0>
20:50:26 <dgilmore> bconoboy: its irrelevant
20:50:36 <masta> hrw: we are using the virtualopensystems that was posted to g+ a while back
20:51:07 <dgilmore> we should keep tabs because people will want to run fedora on their chromebooks, but as things stand we really cant commit to supporting,
20:51:12 <masta> hrw: which is based on the chromeos 2011.12 + linaro hyp patches
20:51:19 <dgilmore> and linaros u-boot tree shouldnt exist
20:51:23 <bconoboy> dgilmore: !?  I'm getting asked this question a half dozen times a day.
20:51:54 <bconoboy> #info linaro uboot 2013.10-rc2 has a chromebook uboot with hyp support (not supported by fedora)
20:52:06 <masta> bconoboy: no!
20:52:10 <masta> that is upstream denx
20:52:12 <hrw> bconoboy: most of users do not want mainline kernel on their chromebook (in current state of it)
20:52:14 <dgilmore> bconoboy: seriously
20:52:20 <bconoboy> masta: is it?
20:52:21 <masta> denx 2013.10-rc2
20:52:25 <bconoboy> ah!
20:52:26 <masta> hehe
20:52:28 <bconoboy> #undo
20:52:28 <zodbot> Removing item from minutes: <MeetBot.items.Info object at 0x27a90d50>
20:52:36 <masta> but we can leave it off the record
20:52:48 <bconoboy> #info denx uboot 2013.10-rc2 has a chromebook uboot with hyp support (not currently supported by fedora though)
20:53:16 <masta> since we do not support, and I'm not 100% certain the patch-set was incorporated, I just remember a few email on the list
20:53:38 <bconoboy> okay, I give up.
20:53:40 <dgilmore> bconoboy: until google or samsung ships the hardware uncrippled we cant give a good user experience, and we should push people to different hardware we can support, though i dont know that hardware exists
20:53:41 <bconoboy> #undo
20:53:42 <zodbot> Removing item from minutes: <MeetBot.items.Info object at 0x2642c210>
20:54:06 <pwhalen> exciting..
20:54:11 <pwhalen> anything else for today folks?
20:54:22 <masta> bconoboy: bconoboy sry for all the topic thrash
20:54:26 <masta> =(
20:54:38 <hrw> time to sleep ;D
20:54:39 <bconoboy> masta: np, if the answer isn't settled yet I don't want to spread misinformation
20:55:14 <bconoboy> pwhalen: seems that's a wrap
20:55:20 <pwhalen> #endmeeting