Kernel Consolidation WG - 2010-07-14 Meeting notes and activity report for last week

Loïc Minier loic.minier at linaro.org
Thu Jul 15 22:37:37 BST 2010


        Hey

 Please find below the minutes from yesterday's Kernel Consolidation
 Working Group meeting and the activity reports from previous week.

 You can find the full meeting notes at
 https://wiki.linaro.org/WorkingGroups/KernelConsolidation/Meetings/2010-07-14

== Agenda ==

 * Action items from last meeting.
  * mwaddel to code review the arm_next.git tree and check that all
    patches are present.
   * [DONE] mwaddel will try the kernel on the versatile board. 
  * jk to report on progress of talking to Grant WRT upstreaming his patches.
   * [CARRY OVER] no response from Grant yet.
  * mwaddel to talk to ARM about device-tree work.
   * [DONE]
  * Jeremy to chat with Nicolas about documenting boot interface.
   * [DONE] some patches already produced.
  * npitre to find a way to expose the list of trees merged in
    arm_next.git via a wiki page.
   * [DONE] documented in the tag instead.
  * jcrigby to talk to slangasek about the list of kernel flavours to
    produce.
   * [CARRY OVER] jcrigby to talk face-to-face at the sprint.
  * amitk to report back the status on linux-omap's patches.
   * [CARRY OVER] amitk absent.
  * jcrigby to determine if versatile is still needed.
   * [CARRY OVER] jcrigby to talk face-to-face at the sprint.
  * lool to check that -fPIE support is currently enabled in the Linaro
    toolchain.
   * [DONE]
     http://lists.linaro.org/pipermail/linaro-dev/2010-July/000150.html
  * npitre and arnd to review ericm's single zImage patches.
   * [DONE] ongoing and doesn't need an action item.
  * ericm to send single zImage patches to russelking after they have
    been reviewed.
   * [DONE] patches are going to the public mailing lists.
  * jcrigby and lool to arrange a uboot specific call for interested
    team members.
   * [CARRY OVER] Arrange at the sprint -
     http://lists.linaro.org/pipermail/linaro-dev/2010-July/000149.html

 * Blueprint status.
 arm-m-using-device-tree-on-arm
 arm-m-kernel-version-alignment
 arm-m-missing-security-features
 arm-m-debugging-with-oprofile
 spec/kernel-maverick-arm-single-zimage
 arm-m-versatile-express-a9-enablement

 * Gathering ideas for face to face discussions / work items for the sprint
 * Personal work plan for the sprint
 * Plenaries
 * Meeting canceled for week of sprint

=== Action Points ===
 * jk to report back on Grant Likely’s response to upstreaming patches
 * ericm to discuss with lool how to get more hardware.

=== Minutes ===
 * lool on vacation today, short status call.
   daverusling asked about the status of the Ubuntu kernel patches in
   the Linaro tree after recent discussion on the mailing list. nico
   explained that they were not to be merged as the benefits of merging
   were small and individual patches could be merged at a later date.
 * mwaddel code reviewed arm_next.git and checked that required patches
   were present. He is checking versatile express with it now.
 * jk emailed Grant Likely to discuss the upstreaming on his patches, no
   response yet.
  * [ACTION] jk to report on progress of talking to Grant WRT
    upstreaming his patches
 * Matt talked to Lorenzo at ARM about device tree status. Based on jk’s
   work patches are not quite ready yet but are close.
 * jk pushed some patches on the boot interface. More work can be done
   face-to-face with nico at the sprint. nico mentioned that
   documentation is already there. Both considered this action done.
 * There was some discussion on where the list of trees merged in to the
   arm_next.git tree should be documented. nico included this
   information directly into a tag in the tree. The question was asked
   if a separate wiki page needed to also convey this information, the
   consensus was not due to the constantly updating nature and nicely
   seeing the documents history.
 * jcrigby has been producing omap and versatile kernel flavours, no
   others have been identified yet. jcrigby will talk to slangasek to
   determine if versatile is still needed and if other flavours should
   be added. This will be done face-to-face at the sprint.
 * eric forwarded some patches to upstream public mailing list for the
   single zImage blueprint. Consensus was that this work spans many
   weeks and isn’t a specific action item but should be monitored.

=== Blueprints ===
 * jk had a bunch of patches around for debugging mapping for device
   tree’s and has been working on them this week. Good progress.
   Discussions with Grant about mainlining device-tree’s proved
   insightful.
 * jk also mentioned that he will be working on imx51 in the near
   future.
 * ericm said that there may be a need for pbuild hacks to allow
   multiple platforms to be built at once to help on the single zImage
   blueprint. Work is ongoing.
 * A lack of hardware is hampering progress at the moment as ericm only
   has one platform.
  * [ACTION] ericm to discuss with lool how to get more hardware.
 * dmart mentioned that the missing security features spec was waiting
   on kees but most items were either nearly complete or completed.
 * dmart went on to inform the group that work on oProfile by willdeacon
   was on track.
 * Work on the versatile express board by mwaddel is ongoing and lots
   have work items have been defined but mwaddel noted that once a
   booting kernel is available most work items will be quick to
   complete.
 * JamieBennett asked the group to fill in their personal goals for the
   sprint at
   https://wiki.linaro.org/Events/2010-07-PlatformSprint/KernelConsolidationWG
   and to suggest plenary topics.
 * There will be no meeting next week due to the face-to-face sprint.

 Activity reports from previous week:

= Arnd Bergmann =
== Getting started ==

 * Built a qemu chroot for cross compiling
 * Research about ARM CPU families
 * Tried to set up Gumstix with uweigand
 * Looked at the kernel for my WM8505 tablet

= Nicolas Pitre =
 * Reviewed some  of Eric Miao's patches on the linux-arm-kernel list.

 * Reviewed Will Deacon's second round of patches to fix asm constraints
   for atomic ops on ARM.

 * Reviewed more of Uwe Kleine-König's Makefile cleanup patches and
   asked him to produce a pull request for RMK.  I also intend to
   pull his patches into the arm-next tree.

 * Had another look at the DT patch series.  Somehow I still am not as
   pleased with them as I would.  Maybe I should take more time to study
   the generic DT docs and possibly come up with more concrete
   suggestions.

 * Discussed with Kees Cook about the issue I was having with the
   stack protector test module.  Seems that it works only on array of
   chars only and not of all types.

 * Tryed to improve design in the OMAP hsmmc driver by discussing
   alternative solutions on the mailing lists where some new development
   were posted for review.  The people involved seem to have agreed to
   change their code.

 * The usual mailing list monitoring and replying.

 * Prepared a Git pull request for the Orion/Kirkwood/Dove stuff I'm
   still maintaining for the next merge window.

= Jeremy Kerr =
== ARM ==

 * device tree: rebased to gcl's new test-devicetree branch and rmk's
   updated master branch
 * device tree: implemented DT-pointer-in-r2 boot interface in ARM
   kernel boot code (r2-direct branch)
 * device tree: implemented DT-pointer-in-r2 boot interface in qemu
   (r2-direct branch)
 * device tree: maintenance on versatile branch
 * chased down versatile timer issues; problem has since been fixed in
   qemu
 * built a simple debug initramfs

= Jamie Bennett =
== General ==

 * Spent much of Monday morning catching up with emails, wiki's and
   various other sources after my vacation. Furious wiki editing and
   tidy-up. Decided to unsubscribe from a few less read mailing lists to
   try to avoid the '' '3000-emails-sitting-in-my-inbox-after-vacation'
   ''syndrome.
 * Took on TLE (Technical Liaison Engineer) role to be ARM's technical
   point of contact. Extra workload at this stage should be manageable
   but will be reassessed at a later date.
 * Usual meetings, kernel consolidation minutes and associated wiki
   pages done.

== Sprint Planning ==

 * Lots of coordinating the Linaro sprint, answering peoples questions,
   arranging hardware to be available, making sure we have enough room
   and working with Anmar to take some tasks from him as he is on
   vacation next week.
 * Arranged with Stephen Doel to supply hardware for the sprint,
   recommended specific items which should arrive by Friday next week.

== Blueprints and Releases ==

 * Started discussions about how we can improve the blueprint/work item
   tracking. For next cycle I will propose some changes to make progress
   on work items much more transparent.
 * Prodded team members individually to ensure they are on top of their
   work items. There is a worrying amount of work still to do this cycle
   and keeping on top of this is essential. Some blueprints need more
   love than others so talked to the managers to ensure everything is
   still on track.
 * The Power Management and Toolchain work groups now have blueprints
   associated with them. These are not tracked with the main release
   burn-downs. Set up launchpad projects for each working group and
   worked with pitti to get the work-item tracker recognising the new
   projects.

== Development ==

 * Spent some time looking at the launchpad-work-item tracker project to
   try to understand how it can be hacked for our purposes. There are a few
   features I will be working on before next cycle to make the Linaro
   project easier to track.
 * Pushed 'members as groups' support to the work item tracker. Now
   projects with members who are actually groups can be tracked allowing
   'meta-projects' to be created to track disparate projects all
   together.
   http://people.canonical.com/~pitti/workitems/maverick/arm-ubuntu.html
   is an example of one such meta-project which is set up to track the 3
   Linaro groups progress.

= Loïc Minier =

 * Reviewed Amit's power management WG plans
 * Tried out UBIFS on my sheevaplug, but couldn't get it to attach to
   mtd2 on boot, would only attach manually after boot
 * Built a maverick kernel for my beagleboard with the workaround for
   LP #591941
 * Looked at ARM's Android patches for RealView; lack hardware to test
   them out, but I could try in QEMU

    Bye,
-- 
Loïc Minier



More information about the Linaro-dev mailing list