Hi Johannes, Johannes Wischert <johanneswi@xxxxxxxxx> wrote: > As some of you might have seen on irc I am currently reviving the > ARM-Port of Haiku. I already got the toolchain to compile (but I > don't > know if it produces usefull code yet ;) ). Sine I have quite some > free time this summer I thought I could continue this porting effort > as a GSoC project offcourse this will be a giant task so the project > goals should be choosen so that they are achievable. > By now I am targeting the beagleboard platform since it is a > relatively cheap and well suported platform. I hope to at least get > the kernel to boot and a serial port driver. > What do you think of this idea? To much/not enough for GSoC ? If you only get the kernel to boot, that might not fill up 3 months; but also have it detect the devices, and even load modules from disk (which requires a few drivers) sounds like it could easily fill up this time. I think this would be a nice and welcome GSoC project. > I want to use the u-boot bootloader already present on the > beagleboard. Since u-boot provides a serial console it could be used > at least during boot time. It can also do elementary cpu > initialization (enabling cache configure RAM etc..) so it should not > be to complicated to get "something" to boot if I create a > ./src/system/boot/platform/u-boot_arm platform or something like that > and use the interfaces provided by u-boot there. If you are using u-boot, you should actually not limit the platform to ARM, but have an u-boot platform, that eventually has an arch directory, similar to how it works for OpenFirmware. U-boot supports many more CPU architectures, and most of the boot loader platform code should be reusable between them. Bye, Axel.