Linux 5.16 Drops Support For MIPS Netlogic SoCs

[tuxd3v]

While I'd like to see how you compiled 68k code into your non-68k kernel, here is a very legacy free unmatched RISCV64 board running Desktop Linux for a really sane desktop experience! https://www.youtube.com/watch?v=Sv4-_a_3BKg ;-)

You are completely right rene,
HiFive should hire you to do that kernel amazing work, because after all they are the ones benefiting of it..
AMd could also help..

[chocolate]

lacek
AlDunsmuir
brad0
René Rebe is doing God's work maintaining his (meta-?)distribution T2 and pointing out the general indifference, if not negligence, toward less popular architectures in the general Linux development, among other things. I personally found one of his videos on Linux's Sparse truly fascinating. So, no trolling here, and his first comment was indeed tongue-in-cheek. User birdie might be right that people tend to become defensive (and therefore resort to the T-word) way too quickly when problems that stem from Linux's lack of automated regression testing are brought to light in good faith, an increasingly unforgivable omission for something as big as Linux. If something like that were present, would dropping legacy code like this older MIPS support code be justified at all?

Anyway, thanks rene.

https://www.youtube.com/user/renerebe/videos

[lacek]

lacek
automated regression testing are brought to light in good faith, an increasingly unforgivable omission for something as big as Linux. If something like that were present, would dropping legacy code like this older MIPS support code be justified at all?

Sure. But I have 2 "buts":
- automated regression testing is surely a necessary development tool. In case of the OS kernel (actually it should be applied to core distro elements, like systemd) this would require massive infrastructure. I am sure there is money for that at Google, perhaps Red Hat.... its is not only hardware. Writing meaningful test is not an entry-level job.
- even given automated regression testing there will be situations where legacy code will be dropped with no functional replacement. They will be just less common. Automated regression testing for each platform would incur some extra cost, and fixing bugs it would still require resources. This would significantly lower the threshold below which one may not care for supporting of a given architecture.

Still This seems to be way to go.