bri3d | # 51 | 2010-01-07, 05:41 | Report

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My kernel is already vastly outdated and never worked anyway, but if anyone wants pointers http://www.natisbad.org/N810/index.html is a fantastic place to start as he seems to have isolated a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.

qole | # 52 | 2010-01-07, 07:45 | Report

bri3d: You made breakthroughs. You found and worked around bugs. Have you reported the bugs at least? Have you submitted your patches? Even though you didn't single-handedly solve all of the problems, you solved a few, and I just don't want those fixes to get lost again, when someone else might be able to push a little farther...

bdogg64 | # 53 | 2010-01-07, 19:48 | Report

I've made some progress trying to get android to boot on the n900. I can tell you what I've done so far.

I took the working 2.6.28 maemo kernel, and use some of the patches from nitdroid and built a new zImage. I download a stock branch of the android source and compiled it, and used the nitdroid setup to create the necessary files android needed. I also added a static build of busybox so I could use it with the bootmenu, because android doesn't have a /usr/sbin/chroot by default.

It boots and I see the Android sign and the init messages, but it gets stuck starting the service. I can see on my sdcard that folders and files are being created and modifed, but I cant get it to boot fully yet. I haven't read into further enough to know how to fix the problem yet, but I think its getting there.

Does anyone else have ideas on how we can get this working?

Thanks

Stskeeps | # 54 | 2010-01-07, 20:42 | Report

bdogg64: I assume that you've flasher-3.5 --enable-rd-mode --set-rd-flags=no-omap-wd,no-ext-wd ?

---

Last edited by Stskeeps; 2010-01-07 at 20:43. Reason: Typo

bdogg64 | # 55 | 2010-01-07, 22:06 | Report

Yep, I have those enabled. I'm gonna go back and look through the documentation more and see if I'm missing anything.

bdogg64 | # 56 | 2010-01-12, 13:59 | Report

Just an update on the progress. I have patched the kernel with the android specific pieces and am able to boot the kernel into Maemo 5. I have been loading the android init through chroot on my microsd card.

After running strace and doing some debugging with device nodes and logs, I'm currently stuck at the dreaded "page flipping not supported" error. At first I thought it would be and easy fix because there is already a patch available, but that patch is for the omap framebuffer video driver (omapfb_main.c) and not the omap2 framebuffer video driver (omapfb-main.c). After some digging, I have yet to find a patch/hack yet to fix the issue.

I think its pretty close if I can get past this. If anyone know where I can find a patch I would greatly appreciate it.

Thanks

twoboxen | # 57 | 2010-01-12, 14:27 | Report

Unfortunately, I don't have a patch that would help you, but I just want to thank you for your effort. There are several of us who are very excited about your work! It sounds like you're pretty close. What branch/version of Android are you working from, btw?

Luke-Jr | # 58 | 2010-01-12, 15:58 | Report

Are you still working with the deprecated 2.6.30-rc8 code or is this with current stuff (new omapfb, etc)?

bdogg64 | # 59 | 2010-01-12, 16:06 | Report

Thanks. If anything its been a good learning experience for me. I am using the eclair branch of Android right now.

bdogg64 | # 60 | 2010-01-12, 16:14 | Report

I'm using kernel 2.6.28 from the maemo kernel source. I started with that one because I knew it would boot and I could patch in some of the android changes based on the work on nitdroid. I suppose since its prior to 2.6.30-rc8, its using the old code. I looked into getting the most recent linux omap dss2 drivers from gitorious but I wanted to see if there was a way to patch this easily before going down that path.