Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Maestro: A Linux-compatible kernel in Rust

blog.lenot.re

Post reply on HN
Print view View original

161–170 of 380 posts

Re: Maestro: A Linux-compatible kernel in Rust

#161

Post by _flux »

cies wrote: post #108
_flux wrote: post #19

Earlier quoted context omitted.

Not even Linux itself is kernel module compatible from version to version, so it would be exceedingly difficult to try to be compatible with it.

Sure, but the benefit from being even partly compatible (same structs with same names, etc -- or maybe some compatibility layer) are great as Linux' devices drivers can then be ported more easily. On one end device drivers in Rust are now possible, OTOH the Meastro kernel. I wonder if there come be a day in my life that I run a non-C-kernel in prod/ on dev laptop.

I hope that there would be architectural improvements possible that would not be realistic for current Linux, and implementing those changes would make also the internals—and thus the kernel module interfaces—look quite a bit different.

Re: Maestro: A Linux-compatible kernel in Rust

#162

Post by cmrdporcupine »

nicklecompte wrote: post #85
cmrdporcupine wrote: post #14

[flagged]

I am not quite sure this discussion applies to this project: "Linux-compatible" and "Unix-like" seems to mean "a POSIX-like kernel" rather than being a direct rewrite of the Linux kernel. The Rust rewrite of coreutils is more problematic because it advertises itself as a drop-in replacement and they are going for complete binary compatibility. I don't think Maestro is actually going for kernel compatibility with Linu…

If I understand this project correctly it's going for syscall compatibility, and can run some subset of Linux binaries already.

Which is cool, but...

In any case, I wish the author luck, and it'd be cool if it went somewhere, but I am disappointed by the choice of license. But that's up to them, in the end.

Re: Maestro: A Linux-compatible kernel in Rust

#163

Post by Qwertious »

jeroenhd wrote: post #78
zelphirkalt wrote: post #40

Earlier quoted context omitted.

Good that I did a CTRL+F before posting. It indeed seems to be a trend to try and supplant GPL software that has served well for decades with hip and trendy probably safer software, that meeeerely has a small catch ... MIT license. I wonder how these projects even attract enough people to be developed. Especially developers of a kernel should be aware of the license side of things, that has protected us all.

These days, I don't think people care about the principles that made Linux a success in the first place. From what I can tell, most people seem to take "open source" as "code I can reuse" rather than some kind of ideology about software freedom. Luckily, most Linux alternatives seem to go absolutely nowhere, so I doubt this project will have much of an impact beyond people thinking "huh, cool project".

>rather than some kind of ideology about software freedom.

And correctly so. "Open source" was explicitly intended to strip the ideology in order to more effectively market the bazaar development model to businesses. That's the difference between Open Source and Free Software.

Re: Maestro: A Linux-compatible kernel in Rust

#164

Post by llenotre »

So many thank to all of you for your support! This project has represented a lot of efforts for me and it means a lot!

Right now the website seems to be pretty slow/down. There is a lot of traffic, which was not expected. I also suspect there might be a DoS attack going on.

I will try to make it work better when I get home! (I am currently at work so I cannot give much attention to it right now)

Sorry for the inconvenience, but glad you appreciate the project!

Re: Maestro: A Linux-compatible kernel in Rust

#165

Post by DaveFlater »

vanderZwan wrote: post #158
DaveFlater wrote: post #133

Earlier quoted context omitted.

That's pure Rust evangelism. As devs we are responsible for what we send out into the world. A schism in the Linux kernel would be bad.

I have trouble not reading this as "forks are bad", which seems to be missing the point of libre software a bit. Am I missing some nuance lost in the brevity of two sentences?

Forks are not inherently good or bad but can seriously muck up an ecosystem if they are not well-justified. Mucking up the Linux ecosystem would be BAD.

Re: Maestro: A Linux-compatible kernel in Rust

#166

Post by cmrdporcupine »

jeroenhd wrote: post #78
zelphirkalt wrote: post #40

Earlier quoted context omitted.

Good that I did a CTRL+F before posting. It indeed seems to be a trend to try and supplant GPL software that has served well for decades with hip and trendy probably safer software, that meeeerely has a small catch ... MIT license. I wonder how these projects even attract enough people to be developed. Especially developers of a kernel should be aware of the license side of things, that has protected us all.

These days, I don't think people care about the principles that made Linux a success in the first place. From what I can tell, most people seem to take "open source" as "code I can reuse" rather than some kind of ideology about software freedom. Luckily, most Linux alternatives seem to go absolutely nowhere, so I doubt this project will have much of an impact beyond people thinking "huh, cool project".

Even though I advocate the GPL, I am not convinced Linux was a success because of the GPL.

It's more a success because of the amazing determination and maintainer skills of Linus Torvalds and the group of enthusiasts that rallied around it in the early days. I have been using Linux (as a user, not a contributor) since pre 1.0 (0.9something on a 486, installed off floppies, fun times.) And it's just amazing where we've gotten to.

Perhaps the GPL played a role in the early day in preventing a parasitical fork by a commercial entity, and by acting as a method of funneling in contributions. Hard to say for sure why Linux won over FreeBSD, though.

Re: Maestro: A Linux-compatible kernel in Rust

#167

Post by snvzz »

px43 wrote: post #159
snvzz wrote: post #105

Earlier quoted context omitted.

If you find this amazing, perhaps you should take a look at seL4, which has formal proofs of correctness, going all the way down to the generated assembly code still satisfying the requirements. It also has a much better overall architecture, the best currently available: A third generation microkernel multiserver system. It provides a protected (with proof of isolation) RTOS with hard realtime, proof of worst case t…

Eh, seL4 has a suite of tools that turn their pile of C and ASM into an obscure intermediate language that has some formally verifiable properties. IMO this is just shifting the compiler problem somewhere else, into a dark corner where no one is looking. I highly doubt that it will ever have a practical use beyond teaching kids in the classroom that formal verification is fun, and maybe nerd-sniping some defense weir…

>(Eh, ) I highly doubt that it will ever have a practical use beyond teaching kids in the classroom that formal verification is fun, and maybe nerd-sniping some defense weirdos to win some obscene DOD contracts.

Uh, perhaps take a look at the seL4 foundation's members[0], who are using it in the wild in very serious scenarios.

You can learn more about them as well as ongoing development work in seL4 Summit[1].

0. https://sel4.systems/Foundation/Membership/home.pml

1. https://sel4.systems/Foundation/Summit/home.pml

Re: Maestro: A Linux-compatible kernel in Rust

#168

Post by lifthrasiir »

px43 wrote: post #159
snvzz wrote: post #105

Earlier quoted context omitted.

If you find this amazing, perhaps you should take a look at seL4, which has formal proofs of correctness, going all the way down to the generated assembly code still satisfying the requirements. It also has a much better overall architecture, the best currently available: A third generation microkernel multiserver system. It provides a protected (with proof of isolation) RTOS with hard realtime, proof of worst case t…

Eh, seL4 has a suite of tools that turn their pile of C and ASM into an obscure intermediate language that has some formally verifiable properties. IMO this is just shifting the compiler problem somewhere else, into a dark corner where no one is looking. I highly doubt that it will ever have a practical use beyond teaching kids in the classroom that formal verification is fun, and maybe nerd-sniping some defense weir…

seL4 actually has an end-to-end proof, which proves that the final compiled binary matches up with the formal specification. There are not many places that bugs can be shifted---probably the largest one at this point is the CPU itself.

Re: Maestro: A Linux-compatible kernel in Rust

#170

Post by diggan »

snvzz wrote: post #109
diggan wrote: post #106

Earlier quoted context omitted.

> Personally, I find yet another monolithic kernel unix clone is not what we need It seems highly irrelevant what you or I need. The author explicitly made the project as a learning experience, not for others. The "Why" is described in the opening paragraph, and makes the goal very clear. And it seems like the author was highly successful, so congratulations author! Great to see people diving headfirst into very comp…

>It seems highly irrelevant what you or I need. There is no need to twist my words into sounding negative. As already explained in the parent, congratulations to them for getting it done, and I believe it does provide value through testing one thing (Rust) while sticking to the very mature and well understood UNIX design.

Sorry if it sounded rough but I don't think I twisted anything. It's fairly common for people on here to post stuff they done as a learning exercise, and also very common for others to then ask "But what is the value?" and "This doesn't seem useful to anyone", which I think could put first-time authors off from HN. I guess I just got a bit tired of it at this point.
Post reply on HN