20:00:17 <pwhalen> #startmeeting Fedora ARM weekly status meeting
20:00:17 <zodbot> Meeting started Wed May 29 20:00:17 2013 UTC.  The chair is pwhalen. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:00:17 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
20:00:17 <pwhalen> #chair pwhalen jonmasters bconoboy ctyler pbrobinson dgilmore dmarlin masta j_dulaney msalter ahs3 agreene jcapik ddd
20:00:17 <zodbot> Current chairs: agreene ahs3 bconoboy ctyler ddd dgilmore dmarlin j_dulaney jcapik jonmasters masta msalter pbrobinson pwhalen
20:00:29 <pwhalen> .fas pwhalen
20:00:29 <zodbot> pwhalen: pwhalen 'Paul Whalen' <pwhalen@redhat.com>
20:00:33 <dmarlin_away> .fas dmarlin
20:00:34 <zodbot> dmarlin_away: dmarlin 'David A. Marlin' <dmarlin@redhat.com>
20:00:36 <bconoboy> .fas blc@
20:00:38 <zodbot> bconoboy: blc '' <blc@redhat.com>
20:00:41 <pwhalen> good afternoon all
20:00:52 * pbrobinson is here
20:01:43 <pwhalen> #topic 0) Status of ACTION items from our previous meeting
20:01:43 <pwhalen> #link http://meetbot.fedoraproject.org/fedora-meeting-1/2013-05-22/fedora-meeting-1.2013-05-22-20.00.html
20:02:06 * pbrobinson wonders if he had any action items
20:02:09 <pwhalen> #info complete -pwhalen to post ks used for highbank F19 Beta RC2 Install, add to the vfad wiki page. Beta ks added to wiki.
20:02:17 <jonmasters> .fas jonmasters
20:02:18 <zodbot> jonmasters: jcm 'Jon Masters' <jonathan@jonmasters.org>
20:02:28 <pwhalen> #info complete -pwhalen to post new image/instructions to the wiki for aarch64 quickstart
20:02:28 <pwhalen> #link https://fedoraproject.org/wiki/Architectures/ARM/AArch64/QuickStart
20:02:43 <pwhalen> pbrobinson, none for you
20:02:49 <pbrobinson> woo!
20:02:53 <dmarlin_away> pwhalen: do we have the image ks files ?
20:03:06 * masta is here
20:03:16 <pwhalen> dmarlin_away, dgilmore posted them to the spins git
20:03:17 <jcapik> .fas jcapik
20:03:20 <zodbot> jcapik: jcapik 'Jaromír Cápík' <jcapik@redhat.com>
20:03:22 <pbrobinson> dmarlin_away: dgilmore pushed them all to the spins kickstarts
20:03:31 <dmarlin_away> thanks
20:03:37 <bconoboy> #info kickstarts used in f19 arm beta are in the spins git repo
20:03:45 <pwhalen> #link https://git.fedorahosted.org/git/spin-kickstarts.git
20:03:53 <jonmasters> excellent
20:04:24 * j_dulaney waves
20:04:49 <pwhalen> anything else from last week?
20:05:33 <pwhalen> #topic 1) Problem packages
20:06:09 <pbrobinson> mostly random side packages. A couple of R-* that previously built
20:06:49 <pbrobinson> eclipse should build but keeps hitting the 24hr limit
20:07:21 <pbrobinson> that's it
20:07:29 <pwhalen> #info eclipse keeps hitting 24hr limit on builders
20:07:40 * j_dulaney may give eclipse a shot locally
20:07:54 <bconoboy> The 24 hour limit should be extended- it was when infrastructure was at seneca
20:08:10 <bconoboy> 48 is more realistic
20:08:19 <pbrobinson> j_dulaney: local makes no difference
20:08:36 <pbrobinson> bconoboy: that needs to go upstream then rather than a random hack
20:08:46 <bconoboy> dgilmore?
20:09:18 <pbrobinson> bconoboy: what I said is basically a quote from dgilmore
20:09:30 <bconoboy> who is going to do it?
20:09:31 <j_dulaney> pbrobinson:  I mean do a basic fedpkg on it, see if it does complete building
20:09:43 <pbrobinson> j_dulaney: it does
20:09:44 <pwhalen> #info from last last week -  systemd's kernel-install calls new-kernel-pkg with wrong parameters (CLOSED) BZ#965897
20:09:55 <bconoboy> pbrobinson: Sure, but if we're all in agreement that is the change, who is going to make the change?
20:10:18 <pbrobinson> j_dulaney: the eclipse team has built it, it built on rawhide and will likely eventually build on f19.... we're right on the cusp
20:11:10 <pbrobinson> bconoboy: the change to make it a config option isn't upstream so its likely to get that upstream and then mainline to agree to change the default
20:11:53 <j_dulaney> Isn't Fedora Infra upstream for Koji, anyway?
20:12:08 <pbrobinson> j_dulaney: no, it's but one user of it
20:12:25 <j_dulaney> Ah
20:12:33 <bconoboy> this is probably a post-f19 issue
20:12:42 <pbrobinson> j_dulaney: redhat, facebook, twitter, cern..... and dozens of others....
20:12:57 <pbrobinson> are all public users of it
20:13:13 <bconoboy> #info koji default timeout of 24 hours is too short for eclipse on current builders
20:13:33 <bconoboy> let's move on- we can discuss how to handle it later
20:13:44 <pbrobinson> agreed
20:13:53 <masta> yes move on
20:14:25 <pwhalen> #topic 2) Kernel Status Update
20:14:46 <pbrobinson> 3.9 is pretty stable... awaiting stuff from jonmasters.
20:14:55 <pbrobinson> 3.10 needs more work
20:15:10 <bconoboy> has 3.10 booted on anything?
20:15:20 <jonmasters> kylem_rht was going to poke 3.10 a little?
20:15:30 <pwhalen> I tried rc3 today on Panda, no go
20:15:41 <kylem_rht> yeah, i'll try to sort it out when i'm in the office tomorrow
20:15:46 <ahs3> upstream 3.10 boots on arndale (arndale needs 3.10, minimum...)
20:15:54 <pbrobinson> we haven't had major config changes
20:16:10 <pbrobinson> from 3.9 -> 3.10
20:16:15 <kylem_rht> i've seen other issues on 3.10 on other architectures, so i'm not terribly surprised it's taking a while to shake things out
20:16:18 <pbrobinson> only the merge of tegra
20:16:26 * j_dulaney is building what will be rc4 right now
20:16:37 <pbrobinson> I can revert that for a scratch build to see if it has any impact
20:16:40 <bconoboy> there's probably some new config value we need to turn on in 3.10 for multiplatform arm to work
20:17:21 <pbrobinson> well that's possible but from the conversations I've had with the linaro guys there shouldn't be
20:18:24 <pwhalen> anything else?
20:18:37 <bconoboy> #info 3.9 good. 3.10 bad.
20:18:44 <pbrobinson> I'd like to hear from jonmasters on his 3.9 kernel action items
20:18:56 <pbrobinson> he's had a couple for the last couple of weeks
20:19:50 <jonmasters> I'm working on the omap dss module ordering/loading issue at the moment
20:20:27 <jonmasters> hope to get an update out in the next couple hours
20:20:33 <pbrobinson> k
20:20:38 <pwhalen> excellent
20:20:46 <pwhalen> #topic 3) Aarch64 Status Update, problem packages
20:20:51 <jonmasters> I walked kyle through the vexpress issues and he wanted to pick that up, so I think I have given him enough to do that
20:20:59 <pwhalen> #undo
20:20:59 <zodbot> Removing item from minutes: <MeetBot.items.Topic object at 0x36782210>
20:21:21 <bconoboy> #info jonmasters working on panda graphics issue
20:21:30 <jonmasters> the CLCD driver stuff is going to be just a dtb problem like I discussed previously but kyle has volunteered to steer it forward
20:21:34 <bconoboy> #info kylem_rht working on versatile express graphics issue
20:21:44 <pbrobinson> OK
20:22:00 <jonmasters> now beyond that I looked at the patches for trimslice and bconoboy is going to try something I suggested
20:22:03 <jonmasters> (for the internal SSD)
20:22:17 <bconoboy> I'm doing nothing of the sort- somebody with an ssd trimslice needs to do so
20:22:31 <jonmasters> well, ok, but I thought you were going to try in the farm internally?
20:22:33 <pbrobinson> jonmasters: is that similar to the timeout issues we saw earlier in the 3.x kernels??
20:22:58 <bconoboy> it would be better if somebody with physical access to the device did so
20:23:04 <jonmasters> pbrobinson: yea, the trimslice SSD controller is just busted :)
20:23:13 <pbrobinson> I have one with the SSD
20:23:20 <bconoboy> can you reliably crash it?
20:23:21 <pbrobinson> it works fine if the SSD is the /home directory
20:23:37 <pbrobinson> but I had issues when I tried to use it for rootfs
20:23:53 * j_dulaney notes that 3.10 build just failed
20:23:54 <jonmasters> pbrobinson: can you run "hdparm -a" on the SSD device?
20:23:55 <bconoboy> We need a good reproducer
20:24:02 * masta too
20:24:04 <pbrobinson> but I prefer it as /home anyway as if root is running on the SD it makes it easy to swap them out
20:24:07 <jonmasters> pbrobinson: I'm interested to know what that says
20:24:12 <masta> trimslice was crashing on ssd, was wondering why
20:24:13 <pbrobinson> jonmasters: sure
20:24:50 <bconoboy> masta: Can you reliably crash yours?
20:25:44 <oatley> .fas oatley
20:25:44 <zodbot> oatley: oatley 'Andrew Oatley-Willis' <andrew.oatley-willis@senecacollege.ca>
20:26:03 <bconoboy> #info If you have a trimslice pro, please test the proscribed hdparm workaround on arm@lists.fp.o
20:26:14 <masta> bconoboy: well I'm not activly trying but I suppose I can monitor the uptime
20:26:17 <bconoboy> #info "Topic: Re: Issues to solve before F19 ARM GA"
20:26:29 <pwhalen> #link http://lists.fedoraproject.org/pipermail/arm/2013-May/006083.html
20:26:37 <bconoboy> tnx pwhalen
20:26:54 <pwhalen> np, next?
20:27:00 <bconoboy> y
20:27:04 <pwhalen> #topic 3) Aarch64 Status Update, problem packages
20:27:21 <bconoboy> According to the nightly reportinator we've built 10335 of 13421 packages
20:27:51 <bconoboy> That's probably a little high- it's counting identical packages with different nvrs, or which there are quite a few
20:28:03 <bconoboy> I'll probably fix that today, so if the count goes down in tomorrow's report, don't panic.
20:28:46 <bconoboy> We could use a hand on failed aarch64 builds
20:29:03 <bconoboy> dmarlin and pwhalen have put together an auto-updating wiki page
20:29:25 <bconoboy> pwhalen: have a link handy?
20:29:27 <dmarlin_away> bconoboy: it's not really auto-updating
20:29:33 <bconoboy> semi-auto-updating :-)
20:29:38 <pwhalen> #link https://fedoraproject.org/wiki/Architectures/ARM/AArch64/Stage4_Problem_Packages
20:29:53 * j_dulaney is still working on GHC
20:30:04 <bconoboy> If you'd like to share in the glory and honor of conquering aarch64, grab a package and go
20:30:19 <j_dulaney> However, I suspect that we're going to need upstream help with it
20:30:24 <bconoboy> (You do this by putting your fas name next to the failed build)
20:30:44 <bconoboy> j_dulaney: jcapik looked into it and reached the same conclusion
20:30:59 <pbrobinson> I pushed fixes for a couple of packages today while I was waiting for my customer's storage morons to unfuck their problems
20:31:21 <bconoboy> pbrobinson: for aarch64?
20:31:32 <pbrobinson> no.... for mips....
20:31:36 <pbrobinson> :-/
20:31:38 <bconoboy> :-P
20:31:45 <jcapik> I've created a feature request upstream
20:31:55 <fossjon> lol pb
20:31:58 <bconoboy> Right now we aren't auto-updating the f19 source rpm repository on arm-temp, so the pool of source rpms is a couple weeks old
20:32:00 <jcapik> the aarch64 is not supported by GHC yet
20:32:12 <pbrobinson> j_dulaney: have you spoken with juhp for GHC?
20:32:15 <bconoboy> #info ghc requires upstream aarch64 support
20:32:23 <jonmasters> dmarlin_away, pwhalen: btw that is a very nice page
20:32:46 <pwhalen> my input was minimal, dmarlins work
20:33:09 <pbrobinson> j_dulaney: if not do so, he's usually around on the arm channel and he's the fedora maintainer and actively assists with problems I've had
20:33:18 <pbrobinson> and helped porting issues etc
20:34:04 <j_dulaney> pbrobinson:  I have not
20:34:11 <bconoboy> That's it unless msalter has something to add (He's doing all the hard stuff)
20:34:16 <j_dulaney> pbrobinson:  I'll ping him
20:34:21 <bconoboy> Hmm no msalter here today
20:34:37 <j_dulaney> jcapik:  Ping in #fedora-arm after the meeting
20:34:45 <jcapik> j_dulaney: ok
20:35:08 <pwhalen> please try the new image and instructions, let me know if you have issue
20:35:13 <pbrobinson> he might have gone home but he's around regularly
20:35:25 <bconoboy> oh pointer, one sec..
20:35:40 <bconoboy> #link Join in, visit http://fedoraproject.org/wiki/Architectures/ARM/AArch64/QuickStart
20:36:04 <pwhalen> ok, next?
20:36:12 <pwhalen> #topic 4) Blockers for F19 GA
20:36:24 <pwhalen> #link http://lists.fedoraproject.org/pipermail/arm/2013-May/006083.html
20:36:34 <pwhalen> that link sums it up nicely, thanks bconoboy
20:36:41 * bconoboy bows
20:36:49 <bconoboy> Looking for volunteers to help with these issues
20:37:03 * pbrobinson needs to reply to that
20:37:08 <bconoboy> The lvm console spam is *really* annoying
20:37:21 <pwhalen> #action pwhalen to look into LVM console spam on first boot, The initial-setup does not appear to run
20:37:27 <pbrobinson> is there a BZ for that?
20:37:34 <pwhalen> agreed, but gone after a reboot
20:37:48 <bconoboy> I'm not aware of a BZ for any of the issues mentioned in the email.
20:37:49 <masta> for TI boards I'd like to propose we do the same as OpenSUSE and use their patches to u-boot so we have ext2 /boot instead of vfat.
20:37:53 <pwhalen> pbrobinson, I havent found one, will open
20:38:20 <pwhalen> #info Restorecon runs on first boot- can we avoid this?
20:38:26 <pbrobinson> does it happen with selinux=0? IE is it related to the labelling issue?
20:38:38 <pbrobinson> the lvm isssue that is
20:38:57 <bconoboy> pbrobinson: I wonder the same thing.  Don't believe anybody has tested booting with selinux=0
20:39:14 <pwhalen> pbrobinson, have yet to investigate, but bconoboy mentioned that earlier as well
20:39:17 <jonmasters> can someone put a link to an fpaste of the lvm error for this meeting's logs if you have one handy?
20:39:27 <pbrobinson> masta: 1) does it work 2) can we use it for the other platforms that need vfat too 3) there's others
20:40:17 <masta> pbrobinson: I'm in the process of making it work, and it does work for the suse guys, and 2) yes it would work across all TI boards. It means we would ntohave to have a 2nd image for TI boards, one image for  everything
20:40:31 <pwhalen> jonmasters, I documented it here, - https://fedoraproject.org/wiki/Architectures/ARM/Quality_Assurance/2013-05-21-VFAD-Fedora_19_Beta_RC2
20:40:31 <pbrobinson> pwhalen: try a new image with selinux=0 and that will give you a answer quickly so you don't waste time on it
20:40:36 <pwhalen> but no fpaste
20:40:42 <dmarlin_away> masta: does it let you just read and write files (MLO, u-boot, etc) as before, but on ext2?
20:40:44 <pwhalen> pbrobinson, sounds good
20:41:11 <pbrobinson> masta: we have more than just TI boards I believe that need vfat
20:41:22 <masta> dmarlin_away: the MLO would be writen to the raw disk at offset 128 KiB using dd
20:41:29 <dmarlin_away> :-(
20:41:37 <masta> so the MLO goes in the raw space before the first partition
20:41:40 <pbrobinson> masta: like any of the cheap imx boards....
20:41:50 <bconoboy> If we were going to pursue anything I'd rather it be combining the omap and non-omap images.  a 20MB vfat partition is not a big deal.
20:41:53 <dmarlin_away> masta: sounds less than desirable
20:42:14 <pbrobinson> and there's others that I can't think off the top of my fried brain without dinner head
20:42:20 <jonmasters> pwhalen: ah, that "activation generator" output is systemd related
20:42:39 <masta> dmarlin_away: I think it's a very good goal to eliminate the omap image...
20:42:48 <pwhalen> jonmasters, yes
20:42:53 <masta> we build just one image, install all the kernels.
20:43:01 <pbrobinson> masta: the omap image is wrongly named. It's not just needed by omap devices
20:43:38 <jonmasters> pwhalen: systemd has "activation generators" that it creates dynamically. In the case of LVM I suspect it's because there is no LVM volume and some kind of error logic is not working right. I suggest we ping agk and point him to some ARM hardware within RH that he can help reproduce this with
20:43:40 <masta> pbrobinson: I was not aware of the other boards, and I guess I dont' own any of them.... will grab a wandboard when we support i.MX6
20:43:48 <pbrobinson> masta: imx, and others I don't have the details to hand need a vfat partition. So while we might remove it for TI based devices we'd still likely need to ship one
20:44:40 <bconoboy> Just a few other things...
20:44:45 <masta> =(
20:44:50 <bconoboy> Nobody at Red Hat is looking at these issues:
20:44:57 <bconoboy> OMAP3 kernel not booting
20:44:59 <pwhalen> jonmasters, found a little about that initially but was sidetracked, will do
20:45:10 <bconoboy> Tegra X is unstable
20:45:26 <bconoboy> - if those don't get volunteers their current condition will be their long term condition
20:45:52 * j_dulaney doesn't have h/w to work on either
20:45:57 <jonmasters> pwhalen: I've pinged agk internally and you can see that too, so I'll let you and Alasdair talk about it/figure out a plan
20:46:09 <pwhalen> jonmasters, thanks
20:46:23 <bconoboy> #info OMAP3 (beagle baord) needs a volunteer
20:46:34 <bconoboy> #info Tegra X stability issue needs a volunteer
20:46:43 <bconoboy> I don't consider either of these blockers for GA
20:46:58 <j_dulaney> bconoboy:  Why not?
20:47:12 <bconoboy> Because they didn't work in F18 either.
20:47:20 <pbrobinson> jonmasters: do you have time to help me debug omap3 with fly swatter at some point?
20:47:23 <j_dulaney> bconoboy:  If we officially support something, and if it doesn't meat criteria, by definition it's a blocker
20:47:42 <bconoboy> j_dulaney: Your if conditional returns false :-)
20:47:43 <j_dulaney> bconoboy:  If it doesn't work, then drop official support
20:47:55 <j_dulaney> Heh
20:48:08 <bconoboy> There is no official support to drop, beagleboard hasn't worked since f17.
20:48:11 <pbrobinson> j_dulaney: we don't block on specific HW really, do you tend to do that for every bit of x86 HW.
20:48:24 <bconoboy> tegra graphics working at all is a brand new phenomenon
20:48:26 <pbrobinson> bconoboy: it has... just not well ;-)
20:48:30 <jonmasters> pbrobinson: yes, let me finish the current outstanding stuff and then let's do omap3
20:48:47 <jonmasters> pbrobinson: first I owe you a working omap4 dss and then an update on mvebu (have not forgotten)
20:48:56 <bconoboy> We have basically 2 weeks to track down and resolve as many of these issues as possible.
20:48:57 <pbrobinson> jonmasters: mvebu is highest prio IMO
20:49:21 <j_dulaney> pbrobinson:  If it is widespread enough, yes, we'll block for specific hardware
20:49:25 <pbrobinson> jonmasters: I might be able to give you some direction post dinner on omap display
20:49:28 <jonmasters> so that might be next week for the omap3. Ok. Well right after I finish this omap debug which should be soon then I'll switch back to mvebu and then we can sync tomorrow on other plans
20:49:38 <bconoboy> #info Clock is ticking: F19 Freeze is June 18
20:50:38 <pbrobinson> anything else?
20:51:14 <jonmasters> dmarlin_away wrote up an excellent page on atomics
20:51:15 <pwhalen> I think that covered it
20:51:34 <pwhalen> #topic 5) Flock Planning
20:51:40 * dmarlin_away will send link to the list
20:51:47 <jonmasters> http://fedoraproject.org/wiki/Architectures/ARM/GCCBuiltInAtomicOperations
20:51:49 <pbrobinson> jonmasters: is he going to be lwn published?
20:52:12 <jonmasters> pbrobinson: I'm suggesting we all post links to that URL and let Jon/Jake know
20:52:15 <bconoboy> If you're going to submit a presentation at Flock you have until Friday.  Friday is the submission deadline.
20:52:17 * jonmasters is putting that link on G+ now
20:52:29 <pwhalen> #link http://fedoraproject.org/wiki/Architectures/ARM/GCCBuiltInAtomicOperations
20:52:45 <pbrobinson> jonmasters: http://www.fpaste.org/15332/13698607/ that's hdparm, want anything else?
20:52:49 <bconoboy> dmarlin: Will you be presenting on gcc atomics at flock? :-)
20:53:30 <bconoboy> #undo
20:53:30 <zodbot> Removing item from minutes: <MeetBot.items.Link object at 0xacba590>
20:53:54 * j_dulaney submitted a talk on arm virt
20:53:58 <bconoboy> #link Fedora FLOCK in August http://fedoraproject.org/wiki/Flock
20:54:13 * pbrobinson submitted a ARM state of the union
20:54:14 <jonmasters> pbrobinson: one sec
20:54:16 <bconoboy> #link Official event site is http://flocktofedora.com/
20:54:36 <bconoboy> #info You have until *this* Friday to submit talks
20:54:43 <pbrobinson> and I need to submit a couple more
20:54:46 * ahs3 needs to submit ideas for a fedora janitors group
20:54:57 <bconoboy> #info pbrobinson to present ARM state of the union
20:55:07 <bconoboy> #info j_dulaney to present on ARM virtualization
20:55:18 <jonmasters> pbrobinson: can you try passing "32" to that hdparm -a" command and see if it takes by running it again without the optional value?
20:55:32 <bconoboy> guys, take it to #fedora-arm
20:55:39 <jonmasters> k
20:55:47 <pwhalen> next?
20:55:48 <bconoboy> So that's 2 ARM talks.  Anybody else?
20:56:04 <j_dulaney> ahs3:  fedora janitors?
20:56:05 <jonmasters> I might reprise the demo I am doing at Red Hat Summit
20:56:16 <jonmasters> that's roughly what I'll submit as a paper to Flock
20:56:22 <bconoboy> #info jonmasters to reprise his rh summit arm demo or similar
20:56:45 <jonmasters> (I've something shiny to show)
20:56:49 <pbrobinson> jonmasters: http://www.fpaste.org/15333/36986100/
20:57:00 <bconoboy> pwhalen: next
20:57:09 * j_dulaney wonders at jonmasters shiny
20:57:18 <bconoboy> now boys...
20:57:19 <pwhalen> #topic 6) Open Floor
20:57:32 <bconoboy> #link David Marlin's guide to gcc atomics http://fedoraproject.org/wiki/Architectures/ARM/GCCBuiltInAtomicOperations
20:58:26 <bconoboy> anything else?
20:58:48 <pwhalen> going once..
20:59:03 <pwhalen> #endmeeting