On Tuesday 7. February 2017 15.01.51 Stefan Monnier wrote:
Not only that: if the machine doesn't run a vanilla Linux kernel, there's a terribly good chance that 3 years down the road, you'll still be stuck with the same outdated kernel.
Right. Even when vendors actually release the corresponding source code and don't drop in binary blobs, due to the phenomenon I call "the Linux speedboat", if they've forked an old kernel and done things their way, and if someone doesn't get on the case immediately, there's the unenviable task of forward-porting that code to whatever it is that the Linux kernel developers happen to like today. If the vendor didn't manage to throw their code aboard the speedboat at the right time, everyone is left floating in the wake.
An example of this that Luke mentioned before was the Skytone Alpha netbook which has an Ingenic SoC that just happens to be the one that there really is no documentation for, although assumptions can be made that it is similar to others that are documented publicly. You can get the sense of how things are by going through the sources for the shipped *2.6* kernel derivative, but it's an exercise in itself to figure out how all that should be redone for today's kernels.
Maybe such forward-porting is not too hard: I actually had a go, not being a kernel hacker, but I didn't sense any genuine interest from anyone who might be better equipped to help such work along. Shinier things take precedence over sustainability and longevity for such people, I guess. Add in weird bootloader and kernel init tricks and you have to be fond of kernel hacking to be bothered. Plus, I don't really see Linux-the-kernel as the future, anyway.
Paul