RE: GP2X hardware platform
>So, about kernel, we just could port 2.4 kernel to 2.6 and there would
>be no problem besides that closed firmware that runs on the other
>processor without MMU (/dev/dualcore).
Not just the second processor as I understand it (also several undocumented
components of the SoC) but that seems a fair assumption. The current 2.4
kernel and user apps also mess a little with the MMU to cache/uncache the
upper 32MB on demand but nothing to really get in the way.
>About bootloader, the same bootloader that boots now could be used
>boot armel kernel, don't you think? As debian-arm is little endian
>stuff and debian-armel is little endian stuff with a new EABI, and the
>kernel is the one that understand about those things. (Is that right
>or i might be missing some point).
I will bow to superior knowledge on that point then. I had in the back of my
mind that U-Boot 1.0.0 had a problem with booting ARMEL (some daft ABI
assumption) but never took my 2.6 hacks far enough for it to be an issue. My
other 2.6 ARMEL projects all use much more recent bootloaders ;).
The only other point is you can't boot a kernel of over 680KB (ish, will
have to check) from the stock bootloader and it has to be flashed to the
NAND. You would need to reflash U-Boot with one of the other variants
(Open2x has one) if you wanted to boot the kernel off the SD/SDHC slot.
If there is genuine interest in porting the kernel to 2.6 (as a precursor to
Debian) then I would be delighted to offer up Open2x's Wiki/Mailing List/SVN
etc. to help with the effort (and my time for what little it is worth) but
as it stands the small number of hands on the project are focused on our
upcoming 2.4 based kernel and firmware release at the moment.
Not to say that people would not get stuck in once there was a concrete
project underway. Like every other project, all it needs are interested
parties with the hardware and the will to make it happen.