View Single Post
Posts: 19 | Thanked: 129 times | Joined on Nov 2009
#51
I'm not exactly sitting on anything exciting: I've got a linux-omap kernel that starts the Android init and then mysteriously crashes in the MMC driver and an Android 2.0 userland that sorta maybe runs under Solca's kernel.

Both require quite a bit of work and time I really don't have (I mostly avoid code at home now - not that it matters much since when I'm not at work I'm at school). I'm also *extremely* frustrated with both linux-omap and the Android kernel at this point - the Android kernel team seems to be doing a decent job but there's no real way to track upstream Linux anymore (since there's no stable tree), so it makes it hard to port to devices which have a seperate project trying to track mainline. Then there's linux-omap, which is just a mess - it's refactored, rewritten, and twisted around 24/7 with seemingly very little regression testing. If you don't believe me check out how many commits are "make xxx device boot again" and so on. I admire the linux-omap project for what they've done, but their scope is way, way big (every OMAP device?!) and I think they're definitely feeling it.

For the userland there aren't any significant modifications, which is why it's mostly broken at the moment - just download a 2.0 Android build environment, check out Solca's NITDroid page for the build scripts, run lunch and make, copy the resulting /system onto the MMC (like old NITDroid) and you're in the business.

My kernel is already vastly outdated and never worked anyway, but if anyone wants pointers http://www.natisbad.org/N810/index.html is a fantastic place to start as he seems to have isolated a linux-omap kernel that works some of the time, which is better than can be said for most linux-omap kernels.
 

The Following 6 Users Say Thank You to bri3d For This Useful Post: