Linus Torvalds just released Linux 5.19 as stable for the newest version of the Linux kernel. He also mentioned this is the first time he released the new Linux kernel from an ARM64 laptop in the form of an Apple MacBook running an AArch64 Apple M2 SoC.
Linux 5.19 brings many new features from initial LoongArch CPU support to continued work on bringing-up AMD Zen 4 CPUs, AMD RDNA3 enablement continuing, more work on Intel DG2/Alchemist, Intel Idle driver support for Alder Lake, initial Raptor Lake P graphics support, Zstd compressed firmware, and some nice performance improvements.
In addition, Torvalds intends to call the next release Linux 6.0, because he’s “starting to worry about getting confused by big numbers again”.
I don’t claim to know English at all but the headline doesn’t look right to me.
Maybe, “Linux 5.19 was released” instead? Or “Linus released kernel 5.19”?
It is correct English – the last word of the title is an adjective here.
I have never given feedback on the correctness of language before. That said, this does not sound right.
I agree that “Linux 5.19 released” sounds just fine to my English speaking brain, especially as a headline.
I think that “released” is an intransitive verb here.
https://www.merriam-webster.com/dictionary/intransitive
“was released” would make it simple past passive tense, something that happened before and thus not news. Essentially you’re having trouble with passive voice sentences.
The fact that he released it using a Macbook with the M2 processor is HUGE. This legitimizes the work the Asahi team is doing, and will have ripple effects benefiting ARM64 Linux development not just on Apple hardware but across the ecosystem. This will get devs interested in the platform, and hopefully speed up overall development and advancement for Linux on AArch64.
I’ve been running it on my M1 Mac since the first public installation script was released, and it still blows my mind just how fast Linux can be on top-notch hardware, even without GPU acceleration. The fact that pre-Alpha, unaccelerated, buggy Asahi on a passively cooled M1 chip is noticeably more performant than the equivalent Arch base on a Ryzen 5 3600 with a Radeon RX GPU is simply insane!
Yup, the M1 is a very well balanced SoC, and the performance cores have a fantastic IPC per watt envelope.
Does anybody know the status of Rust support? I do not see any mention but recall earlier projections that it would be in 5.19.
I did not see it in the 5.20 / 6.0 notes either.
tanishaj,
I don’t know, but it’s something I am interested in too.
While I am very keen on safe languages, I am skeptical of the benefits of using rust alongside a dominantly C code base like linux because it is so heavily dependent on unsafe C interop calls and unsafe C primitives. While there can still be benefit for safe languages within local functions, it doesn’t offer the same scope of benefits. I guess we have to start somewhere though, like kernel modules that are somewhat self contained.
Well, Rome was not built in a day and all that. We agree, you have to start somewhere.
In addition to the additional complexity of adding a second language to learn ( above and beyond C ), I think a reasonable objection to Rust right now would be that it is not available on all platforms that the kernel targets. Rust requires LLVM. A couple of interesting projects are looking to address that but, at the very least, this is layering an experimental language implementation on top of the experimental kernel support.
I think the expectation is though that, at first, Rust would be used to author drivers and modules that are optional to the kernel and somewhat platform specific themselves. This seems like a sane strategy to me and, in my experience, it is often driver code that leads to kernel instability in practice. Where I work, we have a Linux based device used in law enforcement vehicles for which we cannot release an update to the firmware as we are finding that the kernel is crashing in test. We strongly suspect that the source of the problem is a third-party driver ( not shipped with official kernel source ). This is the kind of thing that Rust could really help with in the short to intermediate term.
You are of course correct that partial Rust will do nothing to improve the safety of existing C code in the kernel and that Rust code that calls into C will remain vulnerable. One would at least assume that core kernel C code is well tested both explicitly and by nature of its ultra-wide deployment. Even a little Rust may help in practice.
Baby steps.
tanishaj,
Is there a contract opportunity there?