On Fri, Dec 3, 2010 at 5:37 AM, Bahadir Balban <bbal...@b-labs.com> wrote: > Dawid, > > Thanks for the interest. I would like to announce that we have > succesfully virtualized the Linux kernel over multicore Cortex-A9 > processors. The implementation involves certain novel approaches where > linux applications can switch between kernel and user address spaces > with a neat trick I named as HyperSwitch. This is a great news !
> Unfortunately I haven't been able to release this directly to the open. > I would like to have open source development go hand in hand with a > business model that can self-fund the project. One way to create a self > sustaining model like that is the dual licensed model or open core > licensing. However, my concern is that creating anything self-sustaining > is a challenge on its own, let alone a model where there is open and > non-open options. In conclusion I am not dismissing the open development > option, in fact that would be what I would enjoy the most. However I > need to explore more on how to establish a solid way to take the project > forward to be able to provide also an open option. This sounds perfectly reasonable to me. Anyway, there more code that community can handle in v.3.0, and quite sufficient for development. > One challenge is that when you open up, there is very few people who > would appreciate and make a meaningful understanding or contribution to > the project. In that respect I am considering to do this in a somewhat > organized way where we would promote the use of Codezero and when we > reach a threshold number of requests such as yours (i.e. people start > caring more about this), then I would release it as open source. I think you should really consider promoting this project more. It would be interesting to see slides, presentations and videos on FOSS conferences. OKLabs are really aggressive on this, which is a huge plus : http://www.ok-labs.com/geektv. There is a 23rd annual Chaos Communication Congress in Berlin going on soon (http://events.ccc.de/congress/2006-static/static/2/3/r/23rd_Chaos_Communication_Congress_7c1f.html), I saw recently call for papers... See also what is happening around OsmocomBB project (http://bb.osmocom.org/trac/) as most of the wireless industry geeks are gathered there (although it is a GSM stack on Calypso chipset). I think that this strategy can attract more developers to join and get things moving more quickly. > > If you have other ideas I would be interested to know. > > Going forward, we will add support for Beagleboard, the Pandaboard > (OMAP4 dual core cortex-A9) and possibly Nvidia Tegra2 SoCs. We have > also plans to add support for Ubuntu netbook and Android distributions. > We should be done by all by end of Q1 2011. Personally what I started to look for recently is MIPS support. Especially these seem to be the killers : http://www.linuxnewstoday.org/linux-news-nov-2009-archives/1215-nov-04-2009-linux-news.shtml, overperforming ARM at (much) smaller price. If we could prove that running C0 on one of these for BB and App can be done withouth perf penalty, that could be significant motivation for people to start thinking about the solution, as it will reduce the price and, which is even more important, surface of the SoC. Although I am ARM fan, I will be betting on MIPS in the future. I like taking risks ;). BR, Drasko _______________________________________________ codezero-devel mailing list codezero-devel@lists.l4dev.org http://lists.l4dev.org/mailman/listinfo/codezero-devel_lists.l4dev.org
