Key Points for wider discussion

  • Upstreaming: David is looking at this at the moment. Trying to describe how patches make it into our consolidation tree, and from there to Android. David is working on a documentation page for this.
  • Dropping linux-linaro kernel tree (Nicolas' tree and, for example, what is its value and how does it relate to Android kernel) - relevant thread: Consolidation of kernel trees can take a fair amount of time. Kernel developers are not using linux-linaro tree, but the upstream kernel tree instead, and there are no other consumers of that tree. On the other hand, the value of the linux-linaro tree can be apparent when needing a reasonably stable kernel particularly for "lumpy" upstream patches such as device tree related ones. More opinions are needed in the ongoing list discussion.

Team Highlights

  • ARMHF: working on the documentation work items (see blueprint - for the 11.06 work items). Related to porting Java is now fixed and java packages can be compiled - new obstacle for packages compilation is Haskell (boostrapping of which is being attempted now). Javascript benchmarks are being looked at (trying to do with a basic webkit-based browser - something like midori - to see if there is any performance difference between hard- and soft- float).

    • List of platforms to benchmark: pandaboard, efikas, and possibly a tegra-2 based platform. The benchmarks could also include beagleboards - depending on the success of running the hard float image on those, as well as iMX53 / iMX6 are also considered depending the availability of the boards.
    • Starting to study the LAVA infra and how to incorporate the benchmarking for HF images (debian, ubuntu are going to be available as first target)
  • Boot architecture: SIG weekly meetings have started, trying to nail down a document related to what we care about in boot architecture.
  • Device tree: work has been progressing mostly reviewing the documentation being used - still need to address the reviewers' comments. Also started looking at the initialization order of devices - the init order of kernel modules - with a submitted patch which allows the device drivers to request deferral on probing devices : if the resource cannot be granted immediately then the driver requests the deferral of the probing for a later point in time. Request of a resource actually creates also a resource reference for the driver, which can be used for power management purposes. The patch is being reviewed now, and could simplify some of the performance problems we have now.
  • Java work ( plan progress including a chat with George getting in agreement on the next steps. Regarding the long term plan, it requires some project sponsor as the plan is not sustainable by Linaro alone anymore.

  • QEMU images were tested and bugs found.
  • Need to move to new Versatile Express flavour in ubuntu.
    • Issue that rose during the Versatile Express trial - confirming the relevance of Linaro's work on the kernel side: old code in ARM tree (code not really updated). Documentation is also old. Sending patches to Russell not very easy, since Russell needs patches submitted in a special format.
  • Participation to Dublin event for Ubuntu: focus on discussions related to ARMHF, and server images.


OfficeofCTO/Status/2011-06-28 (last modified 2011-07-04 13:48:30)