Developer Platform Team Meeting

Agenda

  • Actions from last meeting - Steve chasing
  • Engineer progress reports - All team members
  • Work Item Progress and Team Issues discussion - Steve and JamieBennett questions to the team

  • Any other business (AOB)

Past Action items

  • slangasek to review/merge armel-cross-toolchain-base 1.60
  • hrw to merge dpkg-cross 2.6.3 from Debian
  • slangasek to check with tgardner on the timeline for next Ubuntu kernel upload, to see if we can get multiarch linux-libc-dev in

Action Items

  • slangasek to handle merge of dpkg-cross (bug #736154)


Engineers Reports

Your Name <irc nick>

Highlights

  • Short bullet points of work you've done that week which convey progress and highlights which can be used to report on how the team is progressing as a whole.

Plans

  • Your individual plans for the coming week(s).

Issues

  • Your individual plans for the coming week(s).

Marcin Juszkiewicz <hrw>

Highlights

Plans

  • Update Debian cross toolchain package to use gcc-4.6 as default.


Peter Pearse <ppearse>

Highlights

  • Hacked perl now cross builds debian packages
    • Need for arm elf loader was a misapprehension

Plans

  • Tidy up perl cross
  • Take a week off

Issues

  • eclipse work slipping, lack of resource.


John Rigby <jcrigby>

Highlights

  • Pushed another work-in-progress packaged kernel. Configs updated to turn on all the latest greatest features: device tree, lttng, dvfs. Next will be testing to try to get an idea of how broken things are.
  • New u-boot is in bzr on lp waiting for merging. This is based on newly release upstream v2011.03 plus a few patches for device tree and updated mx5 support.
  • Submitted an updated U8500 U-Boot patch series based on feedback I got from previous one plus my own clean up.

Plans

  • Still need to push script cross build patch.


Kunal Goel <kunal>

Highlights

  • Kdelibs and kde workspaces packages uploaded to personal ppa
  • avogadro - due to build time dependency dropped from arm compilation
  • kdeedu - dropped avogadro dep but this also has build time dependency on libOpenGL and libQtOpenGL

Plans

  • Testing of KDE Packages on x86 Platforms
  • Vacation from 9th - 14th April 2011


Frederic Turgis <fturgis>

Highlights

  • Wiki page created to give overview on systemtap + set-up + some real-life example
  • Checked latest testsuite issues vs v1.4:
    • report made to Redhat (no answer yet)
    • report put on wiki page as known issues

Plans

  • check issues: inline assembly, sunrpc and kernel module build-id

Issues

  • pre-empted by TI for urgent issue

Wookey

Highlights

  • Dpkg-cross multiarch upstream release
  • Looked into header locations in multiarch world
  • Made linux-libc-dev co-installable (but steve beat me to it)
  • Fixed klibc to build with new linux-libc-dev
  • Filed expenses for Dallas :-)

Plans

  • Do pkg-config cross-build fixes
  • Upload current multiarch
  • Get back to some xdeb stuff

Issues

  • Nouveau crashes way too often on maverick (twice today)
  • schroot fails to mount bind-mounted /home/wookey in chroots

tgall_foo

Highlights

  • Bugs have been the focus this past week.
  • Bugs in process / addressed:
    • 713104 - in review - autostart the network with dhcp on nano and developer images
    • 746065 - in review - set sane defaults for xscreensaver so it doesn't immediately turn on
    • 728692 - dup of upstream bug under analysis - restart / reboot just logs out
    • 746491 - audio bug
    • 733334 - audio bug
      • Been working through various derivations on my beagle C4 and Xm hardware to get audio working and see what it would take it make it work out of the box on our images. With the beta on ALIP I have not been able to get any sound what so ever. Working with krtaylor on an asound.state file that might be able to be packaged with the images to get a better out of box experience for sound.
  • nano - number of things pushed to my repo, expect a number of merge requests today.

Plans

  • knock off more bugs
  • push changes to support nano diet
    • live helper config changes
    • live helper tooling changes

Issues

  • bug 730912 and 660604 are the two biggest bugs that are impacting my work right now. (broken graphics on beagle C4 and EDID for all platforms)

Avik Sil <aviksil>

Highlights

  • Integrated LTTng into linux-linaro-2.6.38 and sent pull request
  • Verified it on beagle board and panda board. It has been merged to npitre's tree, waiting for the packaged kernel
  • Looked at packaging extra lttng-modules using dkms
  • Trying to understand how dkms is used in other packages like fglrx, lirc-modules-source etc
  • Tried the patch in http://lists.infradead.org/pipermail/linux-arm-kernel/2011-March/045283.html in order to run oprofile on panda. So far not successful, investigating the hang on reading LOCKSTATUS register in CTI module

Plans

  • Generate deb package for extra lttng-modules using dkms
  • Continue investigating oprofile bug #702999 on panda board

Issues

IRC logs

Meeting opened by slangasek at 15:00

https://wiki.linaro.org/Platform/DevPlatform/Meetings/2011-04-06

  • <wookey> hello

  • slangasek waits for hrw, since he has an action to be reviewed

    <slangasek> ok, we'll double back when he comes in :)

    <slangasek> * slangasek to review/merge armel-cross-toolchain-base 1.60

    <slangasek> * hrw to merge dpkg-cross 2.6.3 from Debian

    <slangasek> * slangasek to check with tgardner on the timeline for next Ubuntu kernel upload, to see if we can get multiarch linux-libc-dev in

    <wookey> busy man...

    <slangasek> first one is done - merged but not uploaded, blocked by the second one

    <slangasek> Ubuntu kernel uploads - checked on this, communicated to wookey, and seeking a feature freeze exception for the linux-libc-dev changes: bug #750585

    <slangasek> [TOPIC] Engineering progress reports

Engineering progress reports

  • <slangasek> hrw: hey there! what's the word on dpkg-cross 2.6.3?

    <wookey> slangasek: dpkg-cross 2.6.3 in in debian

    <wookey> so should be OK to pull

    <slangasek> wookey: which is not where it needs to be for me to upload the toolchain packages to Ubuntu :)

    <slangasek> there's an outstanding merge action

    <wookey> ah, OK, I see

    <hrw> slangasek: bug 736154 is still not uploaded

    <slangasek> hrw: my understanding is we need dpkg-cross 2.6.3 merged before I upload any of the toolchain packages; is there any reason I shouldn't claim bug #736154 myself and upload?

    <hrw> slangasek: I can not catch ari-tczew on irc to get info

    <hrw> slangasek: go for it I would say

    <slangasek> ok

    <slangasek> [ACTION] slangasek to handle merge of dpkg-cross (bug #736154)

slangasek to handle merge of dpkg-cross (bug #736154)

  • <slangasek> and then after that's done, all your merge requests are ready for upload, right?

    <hrw> yes

  • slangasek nods

    <hrw> once reviewed and merged

    <slangasek> ok, on to engineering reports

    <slangasek> $ echo $(shuf -e hrw jcrigby dmart wookey slangasek ppearse aviksil tgall_foo kunal fturgis)

    <slangasek> slangasek hrw tgall_foo dmart ppearse aviksil kunal wookey jcrigby fturgis

    <slangasek> rats, drew the short straw again

    <hrw> happens

    <slangasek> last week, started getting bugs filed into the arm porting queue for all packages failing to build in the armel rebuild test

    <slangasek> this week, doko started filing bugs for the x86 build failures as well, so I've done some work triaging those to identify build failures introduced by multiarch

    <slangasek> that list is here:

    <slangasek> [LINK] https://bugs.launchpad.net/ubuntu/+bugs?field.tag=ftbfs+natty+multiarch&field.tags_combinator=ALL - build failures introduced by multiarch

https://bugs.launchpad.net/ubuntu/+bugs?field.tag=ftbfs+natty+multiarch&field.tags_combinator=ALL - build failures introduced by multiarch

  • <wookey> I keep getting 'launchpad timed out' for that URL

    <wookey> tried a few times now

    <slangasek> yes, eventually it loads - and eventually if we keep hammering it, the launchpad team will notice and do something about it ;)

    <wookey> OK, not worth filing a bug over then?

    <slangasek> anyway, there are 37 bugs on there right now

    <hrw> LP + tags shows how young lp is

    <slangasek> wookey: probably not

    <wookey> aha - got it

    <slangasek> the multiarch landing work has calmed down over the past week, so I've been able to get back into other stuff

    <slangasek> y'all should be seeing more of me this week than last ;)

    <hrw> slangasek: reviews?

    <slangasek> hrw: Canonical-only topic, yes?

    <slangasek> oh, you mean merge reviews perhaps :)

    <hrw> slangasek: ;D

  • slangasek switches gears ;)

    <slangasek> yes, reviewing merge requests is right at the top of my list

    <slangasek> toolchain, cross-build fixes, etc

    <slangasek> will be helping to get an armel ppa going for the graphics WG this week as well, so we have somewhere to show our packages

    <slangasek> that's it for me - questions?

    <slangasek> hrw: you're up

    <hrw> ok

    <hrw> sent whole cross toolchain for review

    <hrw> no more update-alternatives!

    <hrw> in Ubuntu it covers 4.4 and 4.5

    <hrw> in Debian it covers 4.4 4.5 4.6 and I gave doko patch for 4.3

    <slangasek> :)

    <slangasek> is Debian keeping 4.3? I thought that was supposed to go away soon

    <hrw> linaro porting jam: provided debdiffs for 3 packages, commented some others

    <hrw> slangasek: it is still in sid

    <slangasek> ok

    <slangasek> I ask because I haven't given him multiarch patches yet, he only asked for 4.4-4.6

    <hrw> dpkg-cross 2.6.3 merged

    <hrw> for next release I would like to drop ubuntu changes from dpkg-cross - upstream does not like them

    <hrw> and they are not documented cleanly

    <hrw> plans: update debian cross toolchain to use 4.6 as base

  • hrw done - questions?

    <slangasek> will you talk to lool about the dpkg-cross changes being dropped?

    <hrw> slangasek: rather with cjwatson as he added them but will catch lool too

    <slangasek> ok

    <slangasek> no other questions from me, thanks

    <wookey> hrw: what did we change?

    <slangasek> tgall_foo:

    <tgall_foo> ok

    <tgall_foo> Bugs have pretty much been the focus over the past week

    <hrw> wookey: config.cache has entries for malloc/realloc needed by lvm2, pkg-config files are changed

    <tgall_foo> Bugs in process / addressed:

    <tgall_foo> 713104 - in review - autostart the network with dhcp on nano and developer images

    <tgall_foo> 746065 - in review - set sane defaults for xscreensaver so it doesn't immediately turn on

    <tgall_foo> 728692 - dup of upstream bug under analysis - restart / reboot just logs out

    <tgall_foo> 746491 - audio bug

    <tgall_foo> 733334 - audio bug

    <tgall_foo> Been working through various derivations on my beagle C4 and Xm hardware to

    <tgall_foo> get audio working and see what it would take it make it work out of the box

    <tgall_foo> on our images.

    <tgall_foo> With the beta on ALIP I have not been able to get any sound what so ever.

    <tgall_foo> Working with krtaylor on an asound.state file that might be able to be packaged

    <tgall_foo> with the images to get a better out of box experience for sound.

    <tgall_foo> nano - number of things pushed to my repo, expect a number of merge requests today.

    <tgall_foo> that's pretty much it

    <hrw> asound.state should be merged into kernel defaults iirc - that was ubuntu policy last time iirc

    <tgall_foo> 728692 it would appear is occuring since shutdown of X isn't happening without error so instead of doing the action to restart or reboot, X gets restarted

    <slangasek> huh, strange

    <tgall_foo> hrw, do you know the relationship between kernel defaults an linaro-media-create ?

    <tgall_foo> a/an/and/

    <tgall_foo> asound.state does appear to be board specific

    <hrw> tgall_foo: by 'kernel defaults' I mean in-kernel config so it will sound outofbox

    <slangasek> tgall_foo: but the kernel knows what board it's running on and can initialize appropriately, can't it?

    <hrw> but that would have to be answered by kernel guys rather

    <slangasek> that's the preferred approach, to be sure

    <tgall_foo> hrw, I don't see how a kernel config will resault in sane alsa state information

    <tgall_foo> slangasek, no .. it's about levels

    <hrw> tgall_foo: s/conifg/source?

    <ppearse> tgall_foo: Used to have a problem with my images where setmixer had zeroed all the levels....

    <tgall_foo> hrw, well the kernel happily sets things to zero and doesn't enable channels like it should

    <tgall_foo> that's what asound.state is for

    <hrw> ok

    <tgall_foo> fun times ahead I suspect :-)

    <tgall_foo> but in short, I'd like to get this working for the release

    <slangasek> great work on getting the images usable out of the box

    <tgall_foo> ppearse, that's an interesting data point .. I'll follow up with you after the meeting on that

    <slangasek> dmart: hi, anything you'd like to report?

    <dmart> The main item is that there's some progress on the VFP coredumps issue

    <dmart> I have a kernel patch for review upstream, and there's also bfd/gdb patches (thanks to uweigand)

    <wookey> so far as I know asound.state is the only way to pre-set levels

    <dmart> npitre: did you have any thoughts on my ptrace patch for this?

    <slangasek> dmart: is that expected to land in Nico's tree this cycle?

  • dmart awaits npitre's reply...

    <dmart> hopefully

    <slangasek> ok

    <dmart> Though we obviously need to merge the bfd/gdb changes to make it useful

  • slangasek nods

    <slangasek> not sure if that's in the set of gdb patches waiting on me for merge into Ubuntu... will have to take a look

    <slangasek> ppearse:

    <dmart> I'll ping for upstream feedback on the kernel patch, though it's been fairly quiet -- so at least nobody really _hates_ it ;)

    <ppearse> Perl cross proposed.

    <ppearse> Using it to creep up on bootstrapping openldap. Hope to finish by the time I go on holiday on Friday.

    <ppearse> That's all this week really.

    <slangasek> ppearse: can you set anything to run in the background while you're away on holiday, so we can get some info about whether multiarch has regressed any of your previously-succeeding builds?

    <ppearse> Possible - I'll let you know tomorrow.

    <ppearse> My chroot fellover so I'm testing quite a few of them anyway......

    <slangasek> heh, ok :)

    <ppearse> To get back to where I was

    <slangasek> thanks

    <slangasek> aviksil:

    <aviksil> OK

    <aviksil> Integrated LTTng into linux-linaro-2.6.38 and sent pull request

    <aviksil> Verified it on beagle board and panda board

    <aviksil> It has been merged to npitre's tree, waiting for the packaged kernel

    <aviksil> Looked at packaging extra lttng-modules using dkms

    <aviksil> trying to understand how dkms is used in other packages like fglrx, lirc-modules-source etc

    <aviksil> tried the patch in http://lists.infradead.org/pipermail/linux-arm-kernel/2011-March/045283.html in order to run oprofile on panda

    <aviksil> so far not successful, investigating the hang on reading LOCKSTATUS register in CTI module

    <aviksil> that's it from me

    <slangasek> do you have any questions on dkms so far, or are you still getting your feet wet?

    <aviksil> slangasek: yes, so far :)

    <slangasek> ok

    <slangasek> let me know if you run into anything I can help with :)

    <slangasek> kunal:

    <kunal> yes

    <aviksil> slangasek: consulting this page: http://tjworld.net/wiki/Linux/Ubuntu/Kernel/BuildDebianDKMSPackages

    <slangasek> aviksil: hey, neat, there is documentation then :)

    <aviksil> slangasek: yes

    <fturgis> aviksil: oprofile patch is to use PMU, not timer based sampling, that's it ?

    <kunal> Kdelibs and kde workspaces packages uploaded to personal ppa

    <kunal> testing is pending as code from mainline

    <aviksil> fturgis: yes, it should use PMU events

    <kunal> avogadro - due to build time dependency on both libopengl and libqtopengl to be dropped from arm compilation

    <kunal> kdeedu - dropped avogadro dep but this also has build time dependency on libOpenGL and libQtOpenGL

    <slangasek> I'm still trying to do a test build of kdeedu here... it fell over overnight because I forgot to enable swap on the panda first

    <slangasek> hopefully I'll be able to comment intelligently on the build soon

    <kunal> ok

    <kunal> thanks

    <kunal> will test kde packages

    <kunal> getting the error

    <kunal> W: GPG error: http://ppa.launchpad.net natty Release: The following signatures couldn't be verified because the public key is not available: NO_PUBKEY FB1B404F70823DCC

    <kunal> graphics WG will test for armel

    <slangasek> sounds like you added the ppa by hand to sources.list; if you use 'add-apt-repository' (from the python-software-properties package), it pulls in the PGP key for you also

    <kunal> i did it by hand

    <wookey> I didn't know there was an 'add-apt-repository'

    <hrw> slangasek: you boot panda with >900MB available?

    <wookey> there are a lot things like that in debian-world

    <kunal> i'll try by add-apt

    <kunal> that's it from my side

    <slangasek> wookey: I often don't use it when I'm in a chroot, since it has a large dependency tree, but yes :)

    <slangasek> hrw: it's booting with the default linaro commandline, which sets mem=463M...

    <wookey> is it python again?

    <slangasek> kunal: thanks!

    <slangasek> wookey: yes, the python-software-properties package is written in python ;D

    <slangasek> wookey: you're up

    <wookey> Finished dpkg-cross fixes after feedback on includes under /usr/lib - upstream now released

    <hrw> slangasek: ro elevator=cfq vram=32M mem=460M@0x80000000 mem=512M@0xA0000000 fixrtc rootwait root=/dev/sda3 console=ttyO2,115200 console=tty1

    <wookey> Looked into header locations in multiarch world (what /usr/lib/*/include/* are actually for, and how many packages)

    <wookey> Made linux-libc-dev co-installable (but steve beat me to it)

    <wookey> Fixed klibc to build with new linux-libc-dev. Still kicking newlib.

    <hrw> slangasek: we need to fix defaults

    <wookey> Did some toolchain packaging for ARM people as fed up of testing with tarball compilers.

    <wookey> Updated installbuilt xdeb branch to revert native/build doc fix, so someone can just hit merge.

    <wookey> Thought some more about where dpkg-cross config.cache info should actually live (and how it works for perl/apache and other non-autofoo systems).

    <wookey> Filed expenses for Dallas before year-end (but still don't have a working windows box)

    <wookey> think that's it

    <ppearse> wookey: DId any of the knowledge you gathered get wikified?

    <wookey> not yet - I haven't reached much in the way of conclusions yet

    <wookey> half-way through an email

    <slangasek> hrw: please file a bug on linaro-image-tools with information about anything that should be changed in the defaults

    <wookey> keep running up against my grep/awk foo limitations

    <wookey> which reminds me I want to grok what steve did for all the *asm* fooage - it looks like exactly what I keep wanting to do

    <wookey> I'll talk to you after slangasek

    <slangasek> wookey: did you see the new bug I reported on dpkg-cross?

    <slangasek> yep, happy to help disseminate archive grepping fu

    <wookey> when? I saw one about xdeb

    <slangasek> (that way next time, I don't have to remember how to do it, I can just ask you ;)

    <slangasek> sorry, I meant xdeb not dpkg-cross

    <wookey> yes. I don;t get the same problem as you (again)

    <slangasek> from the trunk branch?

    <wookey> will look into it some more

    <slangasek> ok, let's talk about that after the meeting as well please

    <wookey> I tried plain natty version

    <wookey> I suspect 'something odd going on' as I can't reproduce your bugs

    <slangasek> yep, we'll bisect our environment later

    <hrw> slangasek: k

    <slangasek> jcrigby:

    <slangasek> (if you're still here :)

    <jcrigby> kernel: pulled latest linux-linaro-2.6.38. Turned on all the latest features

    <jcrigby> trying to test right now ...

    <jcrigby> I don't think it is ready for release, seen lots of bugs on list

    <jcrigby> pushed new u-boot to lp and need to test that before asking someone to merge

    <slangasek> does npitre know your opinion on its releasability?

    <slangasek> (so that we'll hopefully converge on something that is releasable)

    <jcrigby> slangasek, not sure

    <slangasek> ok, please tell him :-)

    <jcrigby> will do, after I have a better idea

    <slangasek> ok

    <jcrigby> of status

    <slangasek> so there's currently no timeline for the next package upload, right?

    <jcrigby> and submitted yet another U8500 u-boot patch series

    <jcrigby> slangasek, not right now, last I saw usb was broken on omap and panda did not boot at all from someone but need to verify that myself

    <jcrigby> thats about it for this week

    <slangasek> ok, thanks

    <slangasek> fturgis:

    <fturgis> Created wiki page gathering all info on systemtap (overview, set-up, real-life example, ...)

    <fturgis> I made wookey aware of this page

    <fturgis> Checked latest testsuite issues vs systemtap v1.4 (and later)

    <fturgis> report made to Redhat (no answer yet)

    <aviksil> fturgis: can you type the link of the wiki page?

    <fturgis> report put on wiki page as "known issues"

    <fturgis> https://wiki.linaro.org/Platform/DevPlatform/systemtap

    <aviksil> fturgis: thanks

    <hrw> slangasek: https://code.launchpad.net/~hrw/linaro-image-tools/more-ram-for-pandaboard/+merge/56602

    <fturgis> My current feeling is that, except kernel module instrumentation issue, issues are not key enough to be worth a new package. We can talk about that later

    <slangasek> [LINK] https://wiki.linaro.org/Platform/DevPlatform/systemtap

https://wiki.linaro.org/Platform/DevPlatform/systemtap

  • <fturgis> got informal approval for Budapest (3 days) so I am booking tonight

    <slangasek> hrw: great :)

    <slangasek> fturgis: great to hear

    <fturgis> And then I was pre-empted this week by customer issues :-(

    <fturgis> that's all

    <hrw> jcrigby: 2.6.38-1207 works fine for me. if you point me to git tree then I can test other kernel on ea1 panda

    <slangasek> fturgis: is a fix available for the kernel module instrumentation, that we *should* do a package update for?

    <jcrigby> hrw, thanks

    <fturgis> yes, Redhat made a fix after finding the issue on powerpc

    <fturgis> maybe I didn't get what you meant

    <slangasek> fturgis: you said the issues aren't key enough to be worth a new package, except the kernel module instrumentation issue - just asking if we have a fix for that issue and if we will be updating the package with that fix

    <slangasek> but we can talk about this later when you're not pre-empted

    <fturgis> ok. I don't how much worth kernel module instrumentation is on SoC

    <fturgis> let's talk later

  • slangasek nods

    <slangasek> [TOPIC] AOB

AOB

  • <slangasek> anyone want to share? good news, bad news?

    <wookey> We broke out the champers this week here - but I can;t tell you why :-)

    <slangasek> heh

    <fturgis> what is a champer ?

    <ppearse> \o/

    <wookey> and xdeb did it's job

    <wookey> champagne

    <ppearse> champers = cambridge for champagne

    <wookey> its (sorry!)

    <wookey> and steve deserves congrats for making multiarch happen at last

    <slangasek> well, we're not out of the woods yet, but things are starting to move in Debian too :)

    <slangasek> should give us some excitement for the next year or so

    <wookey> yeah, it's quite a wide wood

    <hrw> ;D

    <slangasek> #endmeeting


Meeting closed at 16:09

Platform/DevPlatform/Meetings/2011-04-06 (last modified 2011-04-07 08:16:30)