Woke Linus Torvalds rolls his first 4.20, mulls Linux 5.0 effort for 2019

Kernel release candidate adds support for two Chinese CPU families... will these be the final new CPU archs?

The new, improved, and chilled-out Linus Torvalds emitted the first release candidate for Linux kernel 4.20 over the weekend.

He decided not to repeat the jump from 3.19 to 4.0 that happened last year, thus avoiding leaping from 4.19 to 5.0, because “I didn't want to make a pattern of it,” he told the kernel's official development mailing list.

“At this point," the kernel project chieftain added, “I think 5.0 happens next year,” and if all goes well, the Linux 4.20 kernel will be done by December.

His Sunday evening code drop was “70 per cent” driver updates, Torvalds typed, with GPU drivers “looming large as usual” in the “fairly big merge window.”

“The rest is arch[itecture] updates (x86, Arm64, Arm, PowerPC and the new C-SKY architecture), header files, networking, core mm and kernel, and tooling,” he continued.

Chinese outfit C-Sky was acquired by Alibaba earlier this year, having developed an Arm implementation for low-power system-on-chips (SoCs) targeting devices like cameras, DVRs, printers, and media boxes. Kernel stalwart Linaro's Arnd Bergmann last week noted that C-Sky could well be the last new processor architecture added to the kernel. It may be that Arm and RISC-V have rendered all custom CPU designs pointless – chip engineers may as well use either of the pair for their in-house SoCs.

“Both NDS32 and C-Sky are made by companies that also work on RISC-V, and generally speaking RISC-V seems to be killing off any of the minor licensable instruction set projects, just like ARM has mostly killed off the custom vendor-specific instruction sets already. If we add another architecture in the future, it may instead be something like the LLVM bitcode or WebAssembly – who knows?”


Linux kernel's Torvalds: 'I am truly sorry' for my 'unprofessional' rants, I need a break to get help


The other processor architecture added in 4.20-rc1 is an AMD Zen-based data-center processor from Chinese vendor Hygon Dyana.

The 4.20-rc1 also implements the developers' decision in September to remove the National Security Agency's Speck encryption algorithm. Speck and another NSA-developed algorithm called Simon were rejected by ISO back in April, and standards bods accused the agency of bullying tactics.

Phoronix noted that 4.20-rc1 was a huge merge: “By line count addition, it's the largest kernel since last year's Linux 4.13” at over 350,000 new lines of code.

Torvalds said he's also looking at ways to avoid last-minute pull requests: “By Wednesday this week I had gotten a big [sic] frustrated that I kept getting new pull requests when I wanted to really just spend most of the day looking through the ones that deserved a bit of extra attention,” he wrote.

While he does get most pull requests early, “I'm considering trying to make that a more explicit rule that I will literally stop taking new pull requests some time during the second week unless you have a good reason for why it was delayed.” ®


If you feared the newly woke Torvalds would let crap code through into the kernel by being too nice to his fellow programmers, following his month off to reassess his online behavior, perhaps think again. Some were worried his vow to be less shouty and sweary at kernel developers would mean a drop in quality, because he would, er, green light bad code rather than hurt someone's feelings. No, we didn't quite follow the logic on that concern, either.

In any case, with the help of some outgoing mail filters to block any CAPSLOCK-ENABLED OUTBURSTS-ON-AUTOPILOT, and a promise to not be abusive and see a professional for regular therapy, Torvalds has shown it's possible to be his usual direct self albeit without turning the air blue or personal. And it seems it mostly involves using asterisks.

"What makes me *very* unhappy about this is that if I'm right, I think it means that code was literally not tested at all by anybody," he claimed in one email to the kernel mailing list since his return. And in another, the perfectly reasonable:

We do *not* enable new random drivers by default. And we most *definitely* don't do it when they are odd-ball ones that most people have never heard of. Yet the new "BigBen Interactive" driver that was added this merge window did exactly that.

Just don't do it.

Yes, yes, every developer always thinks that _their_ driver is so special and so magically important that it should be enabled by default. But no. When we have thousands of drivers, we don't randomly pick one new driver to be enabled by default just because some developer thinks it is special. It's not.

Please don't do things like this.

So far, so good.

Similar topics

Other stories you might like

Biting the hand that feeds IT © 1998–2021