- Add opt-in Apple Time Machine support
Implement the AAPL negotiation and related Finder, stream,
COPYCHUNK, sparse-file, CHANGE_NOTIFY, and RPC compatibility
required for Time Machine shares, allowing macOS backupd to use
ksmbd for backups
Over my life I have found that with any Apple technology it may work great with anything outside Apple's control... Until it doesn't, usually out of nowhere.
My father is mostly on Apple and tries to use his iPad as a PC, ie with files and "the filesystem", he tries to get out of using extra iCloud for pictures (using Proton, or he asks me about Immich, or he wants to use an external HDD, or use Synology Photos and backup). He's mostly on the iPad + keyboard because it's a nicely managed device update-wise and "what I can install"-wise, virus-wise.
But I recent told him: Just give up. Buy (rent!) the storage, you are too depended on me, there is always something with the non-Apple solutions. Yes, your iTunes library is a mess on the filesystem, just don't look at it through anything else than "Music". Yes, Proton Drive eats 400 GB of your iPad storage now, I don't know how to trim that regularly. I'm sorry, it's either Linux and I can help set some things up, or full Apple. Choose. His current mixed, "trying to get best-of-both-worlds", approach is just pure unexpected pain, every time again.
Edit: I also have an iPad, it does work for me, but where my father is trying to use his iPad as a central, primary device (the hub), for me it is a client (spoke) of my own infra. Apps get access to information and I never use Apple services for anything. So indeed, Immich, Proton, Home Assistant, government and banking apps, etc. My iPad is throw-away: If it dies, it dies... I buy a new one, log into all (self-hosed) services and I'm 100% back up and running. Doing the same for my father is also pain, because his Home Assistant will have issues, or his Immich instance, and it's real work to fix, and he won't tell me for months if it does not work and work piles up. So, I tell him to surrender to Apple... It die a little inside.
>My father (...) tries to use his iPad as a PC (...) He's mostly on the iPad + keyboard because it's a nicely managed device update-wise and "what I can install"-wise, virus-wise. (...) But I recent told him: Just give up
Since he apparently wants a PC, isn't the logical conclusion the opposite? "Just give up. Buy a Windows/Linux laptop, you are too dependent on Apple, there is always something with the Apple solutions."
We made the call to go Apple years and years ago when he was always fighting viruses and virus scanners on Windows. His Macbook was still (hardware-wise) perfectly functional, but does not receive security updates anymore, so he went iPad. Indeed there are more problems than expected. I expected even less IT support from me was needed. Instead he now complains about the lack of control (which I am certainly ok with btw, the complaining I mean.)
The current plan is to put Ubuntu on the Macbook, not sure what to do with the iPad but make it less "primary computer" will be one of the policies to implement indeed. Maybe he will be happy with an external disk and f-spot, and his music nicely sorted into files. Then some generic backup solution to a Synology NAS or something...
Macbooks are great devices, but Linux is a great "fire-and-forget" OS for situations tech-illiterate loved ones. I've done it many times, with much succes and satisfaction (on both ends) without barely any aftercare. All you really need to do is teach them to install those updates and install a major upgrade every few years (or not, Ubuntu will run fine for a decade).
I suspect what we might actually see in that regard is something similar to Minecraft (the Java based one): people were regularly modding it because it was Java based and easy to decomp and know where to inject a mod.
So the Minecraft crew would regularly release updates and obfuscate the code again - most updates were to fix issues/add features, but some were purely for obfuscation. So if you wanted your mod to work it meant not playing any other version.
I'm guessing we'll start seeing more regular updates from software makers; will the updates be fixing bugs or adding features - sure - but I'm guessing that quite a few updates will be simply to obfuscate the code and change addresses around so that any mods that worked before, won't in the new version. Sure an LLM could just decomp it again, but I figure most people won't want to spend the tokens/time it takes for the LLM to actually do the work on that new decomp effort. And with the fact that most people have "auto-update" defaulted to on, I'm foreseeing that apps will get auto-updated likely daily in the near future to combat this.
Years ago, someone once told me "I hope you live in interesting times" ... I don't think this is what they had in mind.
Apple will also use AI to make it at least as painful and as unexpected as before. It's their business model (I have come to believe).
Edit: But good point of course, but I believe it will remain cat and mouse, no idea how Apple internally sees it, whether they really don't care, or actively break stuff (break your ** face tonight).
Reverse engineering is not a magic wand you can wave on everything and expect something good to come out of it. You need to actually know what you are trying to get out of it.
Suggesting that there are these things called computers that are now a lot simpler and reliable would be met with derision because it is a valueless comment
The issue is rarely to have some working setup at a given time, IMHO the real fight is to keep it working thought the updates and ecosystem changes.
That was the same issue with jailbreaking, it gave a lot of freedom, while also being a PITA to deal with at every security fix and system change. Many jailbreakers came to parent's conclusion and just "give up".
Time Machine support has been working on TrueNAS Scale, just via one checkbox on a SMB share config, for years. I thought it was just something in a share config that was always there. Now I wonder how did they do it.
This never really worked for me. It would start a backup, very slowly send over a tiny amount of preliminary data, and then fail after several hours. I'm hoping the in-kernel support will get this working for me, but who knows...
Correct. TrueNAS uses Samba, which has supported this for a while. On TrueNAS it also creates a ZFS snapshot after a successful TimeMachine backup which is really nice for the times MacOS decides to corrupt a TimeMachine volume for no apparent reason.
This really makes me sad. They devoted time and effort to support something as locked and as broken as Apple Time Machine. That thing is a bad choice even within a 100% Apple ecosystem.
Gotta say, using SMB with Apple clients is a massive pain in the ass. I imagine a large part of that is just Microsoft cruft, leaving you to deal with limitations designed around Windows. And you don't really have another option after Apple deprecated AFP.
Apple refuse to pay Microsoft the license for their SMB stack and have dropped open source Samba due to license conflicts with GPL, and so have developed their own implementation (SMBX). This is the reason it is so buggy, and there are loads of articles and posts all over the internet about it. Heres one example from the top of google.
> This release is a record-breaker on all fronts; it brought in 18,765 non-merge commits from 3,021 developers, 807 of whom were first-time kernel developers.
I'm reading that as many new AI assisted code patches. Not everyone does an obviously bad job at submitting AI assisted code.
Broadly it says: you can use AI as much as you like. You have to disclose use of AI. You, as a person, not the AI, are responsible for the quality of Linux kernel submissions.
> You, as a person, not the AI, are responsible for the quality
As someone who doesn't know much about linux kernel work: what does being responsible for the submission look like in practice? Are you expected to find fixes if your change causes unanticipated issues?
I think what it means is "we will judge you based on the code you submit, we don't care if it comes from an AI". Also probably "we will judge you if you obviously don't understand the code you submit".
At least that's how I do it. When someone offers a patch to an open source project, I don't really care if that person has a PhD, or if that person does this in their free time, or if that patch is actually the result of a whole team working together. I judge the patch based on the value it brings to me as a maintainer.
Obviously the code has to be good, but also the interaction with whoever offers the patch has to be smooth: if I have a question, if I need them to confirm/change/test/elaborate on something, I need that to work.
If I feel like I would be faster doing the work myself AND this author is not putting effort into it (e.g. they just copy-paste an AI), I will not review that work. Possibly I will distrust that author. If too many contributions end up wasting my time, I will start limiting the contributions to people I know/trust, etc.
In terms of maintenance, I don't see a difference between getting a contribution from a stranger or from a stranger using an AI.
In terms of philosophy, that's obviously different: One may not want to accept contributions from people using AI because they think that AI is bad for society, for instance. And I would totally understand that. But if the stance (and it seems to be Linus' stance) is that a patch is only judged based on its technical merit, then I don't see why AI-in-the-loop would be an issue.
I've never been involved with Linux kernel work, but in terms of the "You, as a person, not the AI, are responsible for the quality" policy, this to me is a just an "obvious" policy that should be applied to any non-vibe-coded software.
It's just another way of saying, you can't have AI submit things on its own and you can't "blame" AI for anything you submit, you're expected to understand and be able to explain anything you're putting forward, if there is an issue with your submission, doesn't matter at all if you go back to AI to address it, so long as all issues are properly addressed.
The code is either of sufficient quality, or it's not, doesn't matter whether you or AI wrote it, it's your name on it, you're responsible for it and its your reputation on the line.
Basically, you can't just post slop and then say, "Oh, sorry, the AI did that." If the comments are too wordy, or the patch is unnecessary or poorly written or contains errors or doesn't follow the style guide, you submitted it, so that's on you. If you get in the habit of posting bad patches, people may start ignoring you.
Or they don't care whether the code was produced with vim, VScode with an autocomplete tool or AI; they will judge the code based on the quality of the code and the quality of the interaction with whoever submits it?
Nothing says you have to use AI. Just like nothing says you have to use a computer. You could have someone print the code for you, write your patch with a pencil on paper, get a pigeon to send your patch to someone who can then submit it, and it would be judge in the same manner. Granted, it's probably more efficient to use a computer.
On the one hand it is concerning, on the other hand… I was getting quite worried the iPad generation was going to be incapable of developing operating systems. If AI is now capable of maintaining it, that will save us from collapse.
I have this dream that previous gen M chips will be worth more than current gen M chips b/c people want to run local inference on Apple silicon... but they want to use Linux and not OS X.
On the other hand - I just had to send my Asus vivobook s15 to repair, because Linux (most likely) fried the wifi chip.
Pretty disappointed in Qualcomm's promise of upstream support for Snapdragon X Elite turning out to be: lol, not actually so you can run Linux on devices, only so you can maybe boot without most devices working...
Hope things will keep improving despite lack of help from vendors.
> This release is a record-breaker on all fronts; it brought in 18,765 non-merge commits from 3,021 developers, 807 of whom were first-time kernel developers.
hopefully without kernel-slops :P good job everyone btw!
1. They weren't counting LOC though, but commits. It's a totally different metric.
2. This type of stat was always being shared with each new release, giving a general indication of the size of changes. They did not started posting it all of a sudden due to AI.
During peer review a race condition was found in his code. Would he have spotted it himself when writing the code manually? Maybe, maybe not. He's only human, and the AI is only AI.
I think this is yet another case where Linus is leading the pack and showing the industry how to adapt to new technology, and indeed, adopt standard best practices for its application.
I'm in the "AI is great!" optimist camp myself, which is how I felt in the very early days of "this little crappy hobbyist Linux kernel thing that'll never go anywhere important" minix-list flamewars.
Once again, Linus demonstrating that those who understand the code, own the code.
"Showing the industry how to adapt to new technology", as in, don't blindly trust whatever the big probability box committed to your repo but actually perform a good code review with the same standards as human-written code?
I think this says way more about the industry being completely out of touch and less about Linus leading the pack.
Yes, usually when the flock is lost in the wilderness, its leaders need to raise their heads and lead the way. Linus is applying the same great engineering methods he's always had.
Please stop with the "AI assisted development" = "Vibe coding slop" narrative. It comes across as pretty dumb (sorry, I mean "uninformed") nowadays. Please experiment with LLMs, find your way (or don't, but), learn to talk about it in an informed way and don't post obviously inflammatory one liners that are essentially just "AI is bad m'kay".
There are intelligent arguments against using AI, your comment is not referring to any of them.
In short: Linus feels responsible for the quality of the Kernel, he expects the same of other devs. He does not care about the tools used to generate the code. He (or some other human) will call out submitted slop, whether LLM generated or human generated.
It's increasingly the other way around, actually. You are being sloppy if you don't use the tools available to you. Including AI, static code analysis, testing tools, etc. to check if your work is any good. That doesn't mean you have to surrender to the AI tools and blindly generate everything and then commit without testing or any form of validation. That would be sloppy. But then if you manually write your code and did that, that would be sloppy too. Whether you manually type in all your code or have it generated is immaterial. What matters is the work you do to ensure it all works as advertised and is fit to be merged.
The kernel developers are probably a bit further in their thinking then some developers. They've been dealing with many thousands of contributors for decades. Git was invented by Linus Torvalds specifically to deal with ensuring only the best patches make it into the kernel.
Somebody posted a video with Greg Kroah-Hartman last week or so commenting on how he's been dealing with all the AI generated security bug reports. His key point is that AI generated reports are meaningless unless they come with a suggested fix as well. If you are going to generate the report, you might as well generate the fix. And it's a lot easier for him to review if the fix fixes anything worth fixing then to piece together a fix from a lengthy report. The larger the report, the less likely it is to get any attention. If the person contributing can't articulate why the fix is needed or how it works, they just reject the change and toss it back. That strikes me as very pragmatic. Judge a change on its merits. Not on how it was created. A report without a suggested fix is relatively meaningless.
100%. I think there are a lot of people here on HN that are very threatened by LLMs, rightly so. I'm also afraid of my future and my day-to-day work has already changed from "loosing myself in a flow state solving technical puzzles" to "micro-managing LLMs". I don't like it, but hope I will learn to, at some point, or perhaps I find different work. I feel like I don't have a choice right now but to adapt and learn.
The answer to this issue is not to equate any code generated by LLMs to slop. It is just not true. As you say, the latest models often flag things that I would definitely miss, or they simplify my code significantly with a just-released feature. I see benefits to code quality all day from using LLMs.
I'm just saying (not to you but to comments below): Lets keep it constructive, which means: lets keep it real, as in close to truth. Yes vibe coding exists, and it has it's place (ie in prototyping?). But not in the kernel, and also not in my own line of works. LLMs do have a place there though.
There’s such a breadth of slop quality. Like I’ve seen people using ChatGPT with thinking disabled, which is just garbage. But you can also use an LLM to do automated testing of your product, in a real browser window, where it actually sees how it looks via screenshots. That’s insanely useful.
And it can read all of the logs in the backend and browser simultaneously and understand what is causing issues. I’ve used that a few times and it is such a game changer for weird bugs.
The rate at which AI introduces bugs is higher than human ones about 2x more bugs than humans.
Human slop is equally as bad, but this is not what Linus chose in this case, he chose to accept the higher risk of almost 2x as much more bugs using LLMs which lead to the introduction of LLM slop.
I don't care if it is Linus or anyone else, slop is slop human or not, with LLMs it is far worse.
If the human is doing a proper workflow for the AI, I'm not so sure it's big central. Yes, AI needs to be guided strongly, and if it's not, it makes very poor choices/decisions routinely, but that's why you keep an engineer controlling the process.
I force all work to be split up into chunks, properly investigated, spiked, and planned out, and then implemented via TDD (with only system-boundary mocking permitted). It's not perfect, but I've seen regression bugs very much cut so far.
Beyond that, if it's a GUI app, I also require the TDD cycle to require GUI tests with screenshot evidence, and proper assert statements.
Making LLM write code is super susceptible to generating problems imo.
You know it won't be properly reviewed and understood.
Using LLMs to review code is much more useful imo but it can also cause problems due to the natural lazyness of a person.
You wouldn't want the OS you are using to be developed by a dilligent but a low IQ and mediocre skill level programmer.
I'm not saying other people can't possibly use LLMs to generate code effectively but it seems super suspicious from my experience as it triggers laziness and reluctance to put effort into understanding every piece of code.
> You know it won't be properly reviewed and understood
Do you now?
> You wouldn't want the OS you are using to be developed by a dilligent but a low IQ and mediocre skill level programmer.
So you also know that people using LLMs to produce code are diligent but low IQ and mediocre skill?
> it triggers laziness and reluctance to put effort into understanding every piece of code.
Linus has been mostly reviewing code for what... decades? The rest of the contributors have been his LLM in some way. Would you say that he has gotten "reluctant to put effort into understanding every piece of code" he reviews?
Failfs is a pretty cool pseudo-filesystem where every path lookup fails. So if yu place a statically-linked process into failfs with ONLY pre-allocated file-descriptors at process birth, close down unneeded sockets and enforce a default-deny firewall discipline, compromised applications are effectively isolated and unable to access local files or spawn shells during an exploit. An attacker can NOT move adjacently from the exploited process to other parts of the system because to do so they'll need to navigate from root and that would fail since yu set root as failfs and you've got yourself a pure, isolated data-processing pipe that can only interact via the exact file descriptors it was given!!
Unfortunately, yu can't use dynamically-linked apps anyway since failfs would break absolute path resolution loading system libraries.
EDIT::
But I guess yu could put a statically linked app on the socket, wrap it in failfs and gave it proxy requests to a dynamically-linked app process if your app must use something like python.
Is it possible to load the dynamic libraries before fork()ing and somehow prevent the path lookup and library loading? This will open up new possibilities for hardened security models.
That's a fun idea but take python (very familiar)... It's very deeply dynamic and even if yu run imports at the top of your code yu still have many large libs with lazy code execution baked deep inside their internal functions so that assumes yu actually know what code paths the application will hit later. More importantly this goes in the opposite direction of Python's lazy imports thingy.. basically you've got a bomb and yu wouldn't wanna do that. Also, imagine the startup time for large apps. I'd just put a statically-linked binary in front and proxy back to the dynamically-linked process if I can't rewrite the main app in something like Go or Rust.
You explicitly stated as such, I just want to emphasize it again: it only breaks dynamically linked libraries, which - from a stability point of view - is an antipattern anyway.
Almost always youre better off producing a bundle which has everything (excluding the runtime) included... Or just use better languages which categorically don't even have this scenario, eg golang, Java, rust, etc pp
A process which is chrooted into failfs cannot open new files. It is limited to the set of filesystem file descriptors it already has open (presumably set up before the chroot or inherited into the process from some container runtime).
I'd love if chrooting would become easier. In particular a
requirement to have bash or a specific shell is annoying;
imo a simple REPL should suffice already there. I wrote one
that works years ago, need to look into what is all needed
for a chroot (I always used bash for this but I'd also want
to change to my own shell eventually).
We wanted capabilities[1]. Instead we get this over-complicated series of system calls with no design consistency that you have to use very carefully, else they can be bypassed (see the list of caveats on the failfs page).
I'd love to have one that randomly fails, so we could debug e.g. databases and other apps that are supposed to be resilient. I'd call it "havocfs". Doesn't corrupt data, just randomly fails reads/writes/etc. Some software is supposed to handle that gracefully. Would be a great test tool...
Still sucks that it has to be reverse-engineered
My father is mostly on Apple and tries to use his iPad as a PC, ie with files and "the filesystem", he tries to get out of using extra iCloud for pictures (using Proton, or he asks me about Immich, or he wants to use an external HDD, or use Synology Photos and backup). He's mostly on the iPad + keyboard because it's a nicely managed device update-wise and "what I can install"-wise, virus-wise.
But I recent told him: Just give up. Buy (rent!) the storage, you are too depended on me, there is always something with the non-Apple solutions. Yes, your iTunes library is a mess on the filesystem, just don't look at it through anything else than "Music". Yes, Proton Drive eats 400 GB of your iPad storage now, I don't know how to trim that regularly. I'm sorry, it's either Linux and I can help set some things up, or full Apple. Choose. His current mixed, "trying to get best-of-both-worlds", approach is just pure unexpected pain, every time again.
Edit: I also have an iPad, it does work for me, but where my father is trying to use his iPad as a central, primary device (the hub), for me it is a client (spoke) of my own infra. Apps get access to information and I never use Apple services for anything. So indeed, Immich, Proton, Home Assistant, government and banking apps, etc. My iPad is throw-away: If it dies, it dies... I buy a new one, log into all (self-hosed) services and I'm 100% back up and running. Doing the same for my father is also pain, because his Home Assistant will have issues, or his Immich instance, and it's real work to fix, and he won't tell me for months if it does not work and work piles up. So, I tell him to surrender to Apple... It die a little inside.
Since he apparently wants a PC, isn't the logical conclusion the opposite? "Just give up. Buy a Windows/Linux laptop, you are too dependent on Apple, there is always something with the Apple solutions."
The current plan is to put Ubuntu on the Macbook, not sure what to do with the iPad but make it less "primary computer" will be one of the policies to implement indeed. Maybe he will be happy with an external disk and f-spot, and his music nicely sorted into files. Then some generic backup solution to a Synology NAS or something...
So the Minecraft crew would regularly release updates and obfuscate the code again - most updates were to fix issues/add features, but some were purely for obfuscation. So if you wanted your mod to work it meant not playing any other version.
I'm guessing we'll start seeing more regular updates from software makers; will the updates be fixing bugs or adding features - sure - but I'm guessing that quite a few updates will be simply to obfuscate the code and change addresses around so that any mods that worked before, won't in the new version. Sure an LLM could just decomp it again, but I figure most people won't want to spend the tokens/time it takes for the LLM to actually do the work on that new decomp effort. And with the fact that most people have "auto-update" defaulted to on, I'm foreseeing that apps will get auto-updated likely daily in the near future to combat this.
Years ago, someone once told me "I hope you live in interesting times" ... I don't think this is what they had in mind.
Edit: But good point of course, but I believe it will remain cat and mouse, no idea how Apple internally sees it, whether they really don't care, or actively break stuff (break your ** face tonight).
That was the same issue with jailbreaking, it gave a lot of freedom, while also being a PITA to deal with at every security fix and system change. Many jailbreakers came to parent's conclusion and just "give up".
https://www.reddit.com/r/MacOS/comments/18sjnwb/state_of_mac...
I'm reading that as many new AI assisted code patches. Not everyone does an obviously bad job at submitting AI assisted code.
Not a criticism, just an observation.
https://docs.kernel.org/process/coding-assistants.html
Broadly it says: you can use AI as much as you like. You have to disclose use of AI. You, as a person, not the AI, are responsible for the quality of Linux kernel submissions.
[0] https://lwn.net/Articles/1099594/ (user required).
[1] https://lore.kernel.org/lkml/20261009083137.3-1-sashal@kerne...
As someone who doesn't know much about linux kernel work: what does being responsible for the submission look like in practice? Are you expected to find fixes if your change causes unanticipated issues?
At least that's how I do it. When someone offers a patch to an open source project, I don't really care if that person has a PhD, or if that person does this in their free time, or if that patch is actually the result of a whole team working together. I judge the patch based on the value it brings to me as a maintainer.
Obviously the code has to be good, but also the interaction with whoever offers the patch has to be smooth: if I have a question, if I need them to confirm/change/test/elaborate on something, I need that to work.
If I feel like I would be faster doing the work myself AND this author is not putting effort into it (e.g. they just copy-paste an AI), I will not review that work. Possibly I will distrust that author. If too many contributions end up wasting my time, I will start limiting the contributions to people I know/trust, etc.
In terms of maintenance, I don't see a difference between getting a contribution from a stranger or from a stranger using an AI.
In terms of philosophy, that's obviously different: One may not want to accept contributions from people using AI because they think that AI is bad for society, for instance. And I would totally understand that. But if the stance (and it seems to be Linus' stance) is that a patch is only judged based on its technical merit, then I don't see why AI-in-the-loop would be an issue.
It's just another way of saying, you can't have AI submit things on its own and you can't "blame" AI for anything you submit, you're expected to understand and be able to explain anything you're putting forward, if there is an issue with your submission, doesn't matter at all if you go back to AI to address it, so long as all issues are properly addressed.
The code is either of sufficient quality, or it's not, doesn't matter whether you or AI wrote it, it's your name on it, you're responsible for it and its your reputation on the line.
Nothing says you have to use AI. Just like nothing says you have to use a computer. You could have someone print the code for you, write your patch with a pencil on paper, get a pigeon to send your patch to someone who can then submit it, and it would be judge in the same manner. Granted, it's probably more efficient to use a computer.
> Device Tree Updates
> Apple platform gains support for M3 Pro, Max and Ultra SoC
Incredible how fast support for Apple silicon is progressing.
Pretty disappointed in Qualcomm's promise of upstream support for Snapdragon X Elite turning out to be: lol, not actually so you can run Linux on devices, only so you can maybe boot without most devices working...
Hope things will keep improving despite lack of help from vendors.
hopefully without kernel-slops :P good job everyone btw!
Features? Bugfixes? What? Look how many commits we've got!
2. This type of stat was always being shared with each new release, giving a general indication of the size of changes. They did not started posting it all of a sudden due to AI.
Is this satire? :D
During peer review a race condition was found in his code. Would he have spotted it himself when writing the code manually? Maybe, maybe not. He's only human, and the AI is only AI.
I'm in the "AI is great!" optimist camp myself, which is how I felt in the very early days of "this little crappy hobbyist Linux kernel thing that'll never go anywhere important" minix-list flamewars.
Once again, Linus demonstrating that those who understand the code, own the code.
I think this says way more about the industry being completely out of touch and less about Linus leading the pack.
There are intelligent arguments against using AI, your comment is not referring to any of them.
In short: Linus feels responsible for the quality of the Kernel, he expects the same of other devs. He does not care about the tools used to generate the code. He (or some other human) will call out submitted slop, whether LLM generated or human generated.
The kernel developers are probably a bit further in their thinking then some developers. They've been dealing with many thousands of contributors for decades. Git was invented by Linus Torvalds specifically to deal with ensuring only the best patches make it into the kernel.
Somebody posted a video with Greg Kroah-Hartman last week or so commenting on how he's been dealing with all the AI generated security bug reports. His key point is that AI generated reports are meaningless unless they come with a suggested fix as well. If you are going to generate the report, you might as well generate the fix. And it's a lot easier for him to review if the fix fixes anything worth fixing then to piece together a fix from a lengthy report. The larger the report, the less likely it is to get any attention. If the person contributing can't articulate why the fix is needed or how it works, they just reject the change and toss it back. That strikes me as very pragmatic. Judge a change on its merits. Not on how it was created. A report without a suggested fix is relatively meaningless.
The answer to this issue is not to equate any code generated by LLMs to slop. It is just not true. As you say, the latest models often flag things that I would definitely miss, or they simplify my code significantly with a just-released feature. I see benefits to code quality all day from using LLMs.
I'm just saying (not to you but to comments below): Lets keep it constructive, which means: lets keep it real, as in close to truth. Yes vibe coding exists, and it has it's place (ie in prototyping?). But not in the kernel, and also not in my own line of works. LLMs do have a place there though.
And it can read all of the logs in the backend and browser simultaneously and understand what is causing issues. I’ve used that a few times and it is such a game changer for weird bugs.
slop is slop.
Human slop is equally as bad, but this is not what Linus chose in this case, he chose to accept the higher risk of almost 2x as much more bugs using LLMs which lead to the introduction of LLM slop.
I don't care if it is Linus or anyone else, slop is slop human or not, with LLMs it is far worse.
I force all work to be split up into chunks, properly investigated, spiked, and planned out, and then implemented via TDD (with only system-boundary mocking permitted). It's not perfect, but I've seen regression bugs very much cut so far.
Beyond that, if it's a GUI app, I also require the TDD cycle to require GUI tests with screenshot evidence, and proper assert statements.
You know it won't be properly reviewed and understood.
Using LLMs to review code is much more useful imo but it can also cause problems due to the natural lazyness of a person.
You wouldn't want the OS you are using to be developed by a dilligent but a low IQ and mediocre skill level programmer.
I'm not saying other people can't possibly use LLMs to generate code effectively but it seems super suspicious from my experience as it triggers laziness and reluctance to put effort into understanding every piece of code.
Do you now?
> You wouldn't want the OS you are using to be developed by a dilligent but a low IQ and mediocre skill level programmer.
So you also know that people using LLMs to produce code are diligent but low IQ and mediocre skill?
> it triggers laziness and reluctance to put effort into understanding every piece of code.
Linus has been mostly reviewing code for what... decades? The rest of the contributors have been his LLM in some way. Would you say that he has gotten "reluctant to put effort into understanding every piece of code" he reviews?
Unfortunately, yu can't use dynamically-linked apps anyway since failfs would break absolute path resolution loading system libraries.
EDIT:: But I guess yu could put a statically linked app on the socket, wrap it in failfs and gave it proxy requests to a dynamically-linked app process if your app must use something like python.
Almost always youre better off producing a bundle which has everything (excluding the runtime) included... Or just use better languages which categorically don't even have this scenario, eg golang, Java, rust, etc pp
I am the person addressed when someone says "you", not "yu". Why is that person changing how I'm addressed?
It's also confusing because it feels initially like it's some new Linux command or user, like "su".
A process which is chrooted into failfs cannot open new files. It is limited to the set of filesystem file descriptors it already has open (presumably set up before the chroot or inherited into the process from some container runtime).
I'd love if chrooting would become easier. In particular a requirement to have bash or a specific shell is annoying; imo a simple REPL should suffice already there. I wrote one that works years ago, need to look into what is all needed for a chroot (I always used bash for this but I'd also want to change to my own shell eventually).
[1] https://en.wikipedia.org/wiki/Capability-based_security
https://docs.kernel.org/next/filesystems/failfs.html
https://github.com/hakuna-m/wubiuefi