Hello, this is the 2nd report regarding my project. I'll offer a brief summary of how are we doing and then the details, so you don't have to read everything at all.
I have to start saying that over the last two weeks I made a lot of progress on the project. The most important part: Debian ARM image integration with Eucalyptus is now cleared. The work to be done for the next two weeks is pretty much fully coding to achieve what we've committed: supporting Debian ARM images for Eucalyptus. What was done so far -------------------- - I've have identified and mapped what needs to be modified over the eucalyptus source for supporting ARM. I'm puting the details on the next section. I think this is the most crucial part that is now sucessfully achieved. - I started a Project Log[1] on the wiki, so anyone can check and eventyally use the information I've already gathered. The log is being updated constantly. You can also check what I've done and what I'm working on. - We got feeback[2] from Eucalyptus developers and packaging team regarding developers and packaging team regarding the ultimate source for debian/ packaging bits. This was quite important because in the past I've managed to build Eucalyptus from source using the debian/ included in the upstream tarball. However the people in charge told us that they were using a different one, which indeed was different and resulted in different output than I expected. I'm still working on fixing the bits to get a clean build. - I've joined the pkg-eucalyptus team and I've been given commit rights to the repo[2]. 1- http://wiki.debian.org/SummerOfCode2011/BuildWithEucalyptus/ProjectLog 2- http://lists.alioth.debian.org/pipermail/pkg-eucalyptus-maintainers/2011-June/000289.html What I'll be working over the next two weeks -------------------------------------------- - Heavily code on the different bits that I've identified and need to be changed. They are: C code: - node/handlers_kvm.{c,h} - handles nc creation. nc's are node clusters for Eucalyptus, the code is in charge of node creation. It prepares the "machine" that is able to run a compatible image. - util/data.{c,h} - handles instance creation. The code over there implements the functions node/handlers_* use, like instance creation and libvirt's API parameter passing. Since libvirt's parameters for ARM vary from x86's ones, we need to support that. Perl code: - tools/gen_kvm_libvirt_xml - wrapper tool for creating KVM/QEMU libvirt's XML Domain definition. This needs to be heavily changed/updated - redesigned, to support ARM architecture. - Depending on the previous task progress: Modify the Debian packaging bits for Eucalyptus and push them to the pkg-eucalyptus's team repository. This is not crucial for my project's success but I'd like to work with Debian official builds. I've yet to commit the Eucalyptus team version to the debian/ dir. I expect to have fixed the bits to get a successful and clean build, then commit. Other plans ----------- - I expect a slow down on the work pace over the second week due end of semester exams on the week from July 3 to July 9. After that week I'm on vacations and will be fully dedicated to the project. - Code changes to Eucalyptus source will be pushed upstream. I would not like to keep a separate repo for this. I'll talk about this with Steffen. Changes to Debian bits will be pushed to the appropiate repos. Other bits ---------- When the project started I noted it was too vague for some people and it's results/output were unclear. This concerned me. However, while the project started and we got lot of feedback things got much clear and I'd like to take the opportunity to explain it a bit. Currently, developers targeting ARM devices are struggling[1] with having a development environment. There are missing bits that sometimes are not in place and result on hard time for them. Also, due mobile devices growth and vendors using Linux-based[2] operating systems it's a nice opportunity for Debian to keep it's well known fame of: supporting every arch over the sun. How this project fits, well by offering the ability for developers to run ARM-based Debian images over x86/x86_64 CPU's would impact the way they could be more productive and for Debian to be more adopted as development environment. I can say that I'm very enthusiast of this opportunity and I'm certain that the outcome will result in more adoption. Even the Ubuntu people is now pushing more and more efforts for ARM official support, due the facts I've stated. 1- http://lists.debian.org/debian-arm/2011/06/msg00020.html 2- http://events.linuxfoundation.org/events/android-builders-summit/back -- Rudy Godoy http://stone-head.org _______________________________________________ Soc-coordination mailing list [email protected] http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/soc-coordination
