Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
Linux

Greg Kroah-Hartman Chastises Critic, Says Linux Foundation Strongly Supports Kernel Developers (ycombinator.com) 74

It started when Linux blogger Bryan Lunduke complained about how the Linux Foundation was reducing the six-year long-term support (LTS) window for the Linux kernel to two years. Lunduke argued that the Foundation seemed more interested in funding compliance best practices — as well as artificial intelligence and blockchain projects.

In an online discussion, Linux kernel maintainer Greg Kroah-Hartman had this response: Did anyone think to actually ask the developer who is maintaining the long-term support kernel versions why he made that change (back in February?), i.e. me...? No, I guess that would take too much effort, and wouldn't result in such a click-bait headline.

"LTS kernels are no longer supported for 6 years because it turns out no one used them." doesn't have that same fun sound...
In a second comment Kroah-Hartman also clarified that in fact "The amount of resources and other stuff that the Linux Foundation provides to the Linux kernel community has increased over the years, including last year. " Just because new people are brought in with new projects (that the LF member companies want to host) does not mean that somehow less is being given to the kernel community at all. It is not a zero-sum game here at all, that's not how the LF works in any way.

Again, this would have been easy to verify if someone just asked us.

So to repeat, no "abandonment" is happening here at all, the opposite is happening, just like it has for the entirety of the Linux Foundation's existence, support has grown every year.

Thanks to long-time Slashdot reader whoever57 for sharing the news.
This discussion has been archived. No new comments can be posted.

Greg Kroah-Hartman Chastises Critic, Says Linux Foundation Strongly Supports Kernel Developers

Comments Filter:
  • by Ungrounded Lightning ( 62228 ) on Sunday October 08, 2023 @03:15AM (#63909419) Journal

    "LTS kernels are no longer supported for 6 years because it turns out no one used them."

    *I* do! I'd use 'em a lot longer than six years, too, if they were supported.

    Why the HELL should I, and how many thousands or millions of others, have to spend a bunch of time every couple years porting a bunch of machines to a new version of the kernel - and then fixing all the stuff that breaks, and upgrading the machine because the new stuff was designed for faster chips and - when the old kernel and apps would continue to do the job they were set up to do if they just had a few fixes for bugs and exploits patched?

    IMHO the idea that "nobody uses them" is an illusion. Plenty of people use them. But if they're running fine and the patches work, why should they waste their and the maintainers' time spamming the maintainers with "nope, I've got no bugs" reports?

    It's the "Silent Majority" problem. How do you judge the size of the population that's NOT complaininig?

    • Re: (Score:2, Informative)

      Why the HELL should I

      You're getting shit for free. You have no right to make any demands.

      • by alantus ( 882150 )
        In that case, the justification is not that nobody uses kernel versions older than 2 years, it's just that users don't pay for it, so they have no rights.
        • We in general have come to accept certain things about the english language, one of them is that it's not to be read like an autist solving a logic puzzle. No one actually thinks that "no one used them" means that not a single person on the earth is running a 6 year old LTS build. And if you are going to claim that you thought that making my post incorrect then you have spectacularly missed the point.

          • by alantus ( 882150 )
            Ok, so we're back to the first reason then, it's not about money anymore.
            I wrote "nobody uses kernels older than 2 years" exactly the same way as is written in the article, I never implied that no single person uses it, yet you took the "autist solving a puzzle" interpretation.
            The truth is nobody knows how many people use older kernels, everyone is just making assumptions. I can say for sure that I use them.
        • In that case, the justification is not that nobody uses kernel versions older than 2 years, it's just that users don't pay for it, so they have no rights.

          I disagree!

          The users have complete rights under GPL to take the code, modify it, and use it as they wish as long as they make their changes available under the same agreement.

          They have the complete right to patch and modify the code as they wish. They have more rights than anyone running Solaris or HPUX or Windows-XP or whatever old system you wish to name. You are confusing their desire to have a free ride from others with having the rights to do with the code as they wish.

          If you want someone else to

          • by alantus ( 882150 )
            You completely missed the point, which is that the real reason is not that nobody uses older kernel versions, that's just an excuse based on an false assumption.
      • by dfghjk ( 711126 )

        He's not making demands, he's correcting a lie.

    • Re: (Score:1, Flamebait)

      Why the HELL should I

      Nothing worse than entitled dickheads who act like they're the ones being burdened by things they get for free.

      Yeah. YOU'RE the one who is so burdened. How dare they provide you with millions of person-hours of work and not also wipe your arse for you too?

    • by vadim_t ( 324782 ) on Sunday October 08, 2023 @03:30AM (#63909431) Homepage

      If you want your particular concerns to be taken into account, you have to be visible in some way.

      Show up to meetings/events, complain about problems, just reply "Thank you, this fixes a problem I had" to updates to make it clear you're there, pay for paid support, or for the lazy way, enable telemetry.

      If nobody knows you're there, you're indistinguishable from not existing. This is especially the case in open source where you can download some software, and use it daily for years without leaving any trace of that anywhere.

    • Re: (Score:2, Insightful)

      by derplord ( 7203610 )

      > Why the HELL should I, and how many thousands or millions of others, have to spend a bunch of time every couple years porting a bunch of machines to a new version of the kernel
      Then hire someone to maintain them.

      Why would GKH maintain this stuff for you, for free, exactly? Because it's open source? Please.

      • Re: (Score:2, Insightful)

        by Anonymous Coward

        That implies the reason is not the claimed "no one used them" but rather "nobody was paying me enough to maintain them", which seems like something the Linux Foundation might be able to help solve.

        Congratulations on missing the OP's whole point.

      • Re: (Score:3, Informative)

        by dfghjk ( 711126 )

        "Why would GKH maintain this stuff for you, for free, exactly? Because it's open source? Please."

        GKH isn't a volunteer, he's paid very 300K a year to work on "open source" specifically to provide this support. So "why would GKH maintain this stuff for you, for free, exactly?" BECAUSE HE IS PAID TO. Please.

        • I guess if he can say, I'm not doing that anymore, and he still gets paid, I guess he wasn't paid for that precisely...?
    • by serviscope_minor ( 664417 ) on Sunday October 08, 2023 @05:39AM (#63909579) Journal

      *I* do! I'd use 'em a lot longer than six years, too, if they were supported.

      It's free, so if it breaks then you get to keep both halves.

      Question is though, why are you using very old kernels (genuine question, I'm not being snarky), and are you using one of the vendor supplied ones? Ubuntu, dead rat etc do LTS on their kernels for quite a while. 10 years if you pay 'em.

      and upgrading the machine because the new stuff was designed for faster chips and - when the old kernel and apps would continue to do the job they were set up to do if they just had a few fixes for bugs and exploits patched?

      Specifically what? You can run old programs on new kernels if you need to. In addition, new OSs aren't exactly sluggish on old hardware. I'm typing this post on a thirteen year old laptop running the latest Linux mint, and it's pretty good. My substantially slower raspberry pi 3 runs the latest picade just fine, too.

      No snark, but specifically what's your usecase?

      • by lmfr ( 567586 )
        In my case, audio over HDMI on my htpc breaks with 5.x, works great with 4.18
    • > Why the HELL should I, and how many thousands or millions of others, have to spend a bunch of time every couple years porting a bunch of machines to a new version of the kernel

      Because you're a part of this world, and building something means maintaining it until it stops being used.
      This create and forget culture is part of why bridges are crumbling all over the western world.

      • by dfghjk ( 711126 )

        So operating systems crumble with age? The thing with analogies is that they need to be applicable.

        • Re: (Score:3, Interesting)

          by Anonymous Coward

          They do indeed "crumble". New exploits get taken advantage of, new hardware to work with/work around etc. Maintenance is quite expensive, hence the desire for many to go to throw away culture. No need for the expensive maintenance costs that may or may not be worth the value. A new thing on the other hand has a defined entry cost.

        • Why do you want updates and support if they don't crumble?

    • by ljw1004 ( 764174 )

      It's the "Silent Majority" problem. How do you judge the size of the population that's NOT complaininig?

      Telemetry is a tried and successful approach...!

      • It's the "Silent Majority" problem. How do you judge the size of the population that's NOT complaininig?

        Telemetry is a tried and successful approach...!

        With security in mind, no one enables telemetry. The last thing I want to be is targetted by someone who knows exactly what I'm running.

    • by Junta ( 36770 )

      Question is what is your total software picture? Sure, we are fixated on the kernel, but an OS is so much more than the kernel. How many components that you use have an upstream LTS policy at all, let alone one that covers 6 years? How many of those are just as functionally and security critical as the kernel when you get down to it?

      Practically speaking, having your kernel and only your kernel supported is insufficient anyway. Hence why folks either:
      -Use a distribution that extends LTS over projects usi

    • Comment removed (Score:4, Insightful)

      by account_deleted ( 4530225 ) on Sunday October 08, 2023 @08:49AM (#63909747)
      Comment removed based on user account deletion
    • by BadDreamer ( 196188 ) on Sunday October 08, 2023 @09:47AM (#63909815) Homepage

      So your use case is somehow that you want to keep patching an old kernel, on top of which you have an old userspace which does not get patches in that way, and use it for... what?

      If you need security updates, you need those for your whole chain. And you won't get that. You will be forced to update userspace anyway, meaning you will need to deal with all the breakage no matter what. The kernel won't give you any of that, it's remarkably backwards compatible.

      And if you're keeping a stable userspace because you don't need security updates, then you don't need kernel updates either. As you note, if they run, they run.

      Nothing about what you state makes any sense. I call bullshit on this whole scenario clearly invented to cast shade and sow discord.

    • The fact that you either are too stupid to understand that he doesn't literally mean zero people or are intentionally being dense just reinforces the fact that you are nobody. So if you use it then it is certainly correct to say that nobody uses it.
    • by thegarbz ( 1787294 ) on Sunday October 08, 2023 @11:39AM (#63909985)

      It's the "Silent Majority" problem.

      No it's an imaginary majority problem; the idea that just because you do something it is popular. It's incorrect.

      • It's the "Silent Majority" problem.

        No it's an imaginary majority problem;

        Precisely. "Silent Majority" refers to a particular political episode in the Vietnam Conflict era, where large numbers of the population were protesting the war - to the point of marches that filled multi-lane main streets for miles - and the administration claimed that the population was really mainly FOR the war but weren't protesting because the government was doing what they wanted (and HAD elected them earlier).

        My point was not

    • Kernel upgrades are minutes a machine, can be automated, and don't break userspace. And the box should be being rebooted more often than "every couple years" anyway.

      This is a non-thing.

      • ksplice means he doesn't have to boot just because you think he should boot.

        • ksplice reduces, not eliminates, the need to reboot after changing the kernel. Patches that change data structures still require reboots.

          Also, he's never upgrading libc6?

          And with automation, who's bothering with that anymore anyway? Patching-without-reboot was a "I have one physical server running the show and it can't go down" era thing.

          There's no Venn diagram for "I'm doing such bleeding edge stuff that kernel updates require me to re-validate my application stack" and "I'm also doing all that with no a

  • Microsoft mostly phased out its LTSC range by severely restricting its availability, support length and not providing a Windows 11 version. Unless you are a micro-controller with a few kb burned in rom you can expect to need constant updates in this internet of things ecosystem. Most websites don't even support Firefox ESR either because they expect the latest version of your web browser at all times. We have established tools to make sure your system is up to date (apt/zypper/dnf/pacman), so no need to wo
    • by Anonymous Coward

      Yeah if everything is moving to "as a service" and eventually expensive, those mainframe stuff start to look attractive especially if you actually have zillions of lines of code.

      Because at least you don't have to keep changing, retesting and debugging your old code every 18 months. Maybe you might have to do that every 18 years (or someone else has to do it after you've retired): https://www.longpelaexpertise.... [longpelaexpertise.com.au]

      Take our CICS example above. IBM stated that all new CICS functions would only be command level from CICS/VS 1.4 in 1978. Smart CICS users would have seen the writing on the wall, and started planning their migrations. The last CICS to support macros was CICS/MVS 2.1.2, released in 1987. So that's 9 years. Even nicer, IBM continued to support CICS/MVS 2.1.2 until around 1996: Another 9. So users had 18 years to convert over.

      I suspect many organizations would be happy to remain on Windows 7 for decades if it was suppor

      • by jmccue ( 834797 )
        Even though you are AC, if I had points I would mod this up. FWIW, RHEL, owned by IBM, does keep patching old kernels for 10 years, it is just no one can get the patches any more unless you pay.
    • by dfghjk ( 711126 )

      LOL wait until you graduate and learn how the real world works.

      "We have established tools to make sure your system is up to date..."

      You have, have you?

    • >"Most websites don't even support Firefox ESR either because they expect the latest version of your web browser at all times."

      That isn't my experience at all, using ESR. In fact, it is almost unheard of that even using an OLDER ESR gets blocked, much less a CURRENT version of ESR.

      Having such an experience is just a crappy filter that maybe an extremely tiny number of crappy websites use. "Oh, you are using 115, that is not current." I have run across this only once, a few years ago, and I actually co

      • Sorry, messed up the list, try 2:

        ESR 115.0.0
        ESR 115.0.1
        ESR 115.0.2
        ESR 115.0.3
        ESR 115.1.0
        ESR 115.2.0
        ESR 115.2.1
        ESR 115.3.0
        ESR 115.3.1

        That is in 3 months, so far.

      • If a website doesn't work on Firefox (under ANY OS), it means it is not following open standards

        Even in Firefox under iOS? Say a web application's script runs feature detection while it starts and displays an error message if a needed web platform API is not present. Tapping "Technical details" below the error message reveals a link to caniuse.com and to the article "Progress Delayed Is Progress Denied" by Alex Russell [infrequently.org]. It could mean that Apple WebKit, the engine of Firefox for iOS and all other web browsers on iOS, is lagging behind the open standards by several years.

        • Well, there is that- any "third party" browser under iOS is an anomaly, courtesy of Apple.

          My notation does hold under all other mobile and all desktop, though.

  • He did not do this because nobody used them for that long, quite the opposite, and if anything, billion dollar companies been freeloading to mainline developers detriment.

    Up to 2 billion people use LTS Linux kernels because that is what Android Common Kernel is based upon and what the modern Android GKI images are based upon too. Between Android, ChromeOS, Debian and Ubuntu LTS, it becomes very fair to say the lions share of real-world users likes them a lot more than mainline. Heck, even specialised sit
    • The solution to this is simple: have phones be like PCs one can install OS upgrades and updates from Google directly, or from third parties if they'd prefere; have specialized hardware have drivers downloadable, upgradeable and updateable from the manufacturer's website; have the interfaces for all this be standardized.

      There, done, now the BS of phones only receiving OS updates for a few years, if at all, unless they're rooteable, and unless they're famous enough to get someone else to develop new customize

      • Indeed, there are many solutions. For sure all of them involve the people who want it (i.e. large companies) not freeloading.

      • by tlhIngan ( 30335 )

        have specialized hardware have drivers downloadable, upgradeable and updateable from the manufacturer's website; have the interfaces for all this be standardized.

        That only works if the driver API is stable. Linux has never promised driver API stability ever. Thus all drivers need to follow the kernel version in use, because a driver that builds using x.y.z version of the kernel is not guaranteed to work on x.y.(z+1) of the kernel.

        Traditionally the kernel and drivers have moved together - and often times by

    • > https://youtu.be/HeeoTE9jLjM?t... [youtu.be]

      The Kroah man himself tangentially explaining why LTS kernels don't work -- no one who matters uses them correctly. (mainly because a bug is a bug is a bug...)
      Fun takeaway: many phones that shipped with kernel 4.14 probably have broken spinlocks.

    • by Junta ( 36770 )

      Debian is a fair point, they do align with LTS kernels. However, notably their security team discontinues maintenance on a release after 3 years, and they explicitly recommend moving on before that three years is up. There is a LTS to get to 5 years, but for whatever reason the security team recommends you not bank on that.

      Ubuntu LTS *technically* has LTS kernel, as inherited from Debian. I'm not sure they care. They will push LTS users to non-LTS "HWE" kernels. Also, when new LTS comes along, they agg

    • Up to 2 billion people use LTS Linux kernels

      He didn't claim that people aren't using LTSe" (meaning f Linux kernels. He claimed that "no onew people, not literally no one) is using them for that duration.

  • by Mr. Dollar Ton ( 5495648 ) on Sunday October 08, 2023 @04:20AM (#63909473)

    I, for one, use 2.6 kernels and I donate to the foundation. But then, I also support my stuff myself, so...

    • I, for one, use 2.6 kernels and I donate to the foundation

      I didn't realize they took Confederate dollars...

    • OK, but why? Genuinely curious. Some retrocomp thing or are you running a current desktop with an ancient kernel?

      • Long-running scientific experiment. Super old hardware, matching firmware, would rather not even stop it, has to run for 4 more years.

        • OK as a nerd you have me really interested now!

          Contemporary hardware from the 2.6 kernel (so 2006 or so?).

          What does the experiment do?

          • The experiment is investigating the origin of some gamma rays that allegedly appear from lightning in the atmosphere. That is, it has an arm that detects the lightning, and another arm that detects the gamma rays over an area. Then it is trying to find if the two originate inside the same time window and then use the data to figure out what caused the gamma emission. It is hard, because unless the storm happens above your detector, you're unlikely to catch any gammas and besides you can't really tell where

            • sounds pretty fun and a perfectly good reason to use an ancient kernel.

              I assume the machine never gets updates and isn't on the internet anyway.

              A bunch of really custom hardware in those data collection boards I'll bet? Sounds like an interesting experiment.

  • Earlier this past summer, IBM complained that their release of CentOS sources supported lots of freeloaders, and so chose to stop.

    So now last month Greg Kroah-Hartman (who works for the Linux Foundation, as far as I can tell) says he's only going to maintain old kernels for two years, going forward. Has IBM been freeloading off his hard work all this time? RHEL 8 appears to be running kernel 4.18 right now.

    • by transwarp ( 900569 ) on Sunday October 08, 2023 @05:24AM (#63909575)
      No. Red Hat (nor all the other commercial distros) don't use the official LTS kernels. 4.18 isn't one, 4.14 and 4.19 are, not that Red Hat kernel version numbers match upstream anyway. "Nobody uses them" should be a mea culpa that they launched the program without consulting the expected users, but that doesn't seem to be the lesson anyone learned.
    • by dfghjk ( 711126 )

      "Has IBM been freeloading off his hard work all this time?"

      No, because he's a paid employee. His "hard work" is not his property, it's his employer's.

  • By far, most of the board, if not all, of the Linux Foundation Board are all employed by Large Companies

    https://www.linuxfoundation.org/about/leadership

    And very little of its funding goes directly towards Linux itself. Plus I could not find who the large donators are after googling for a while. But I have seen in the past most of the finds comes from companies like Microsoft, IBM and Oracle. You think they donate because they are nice ? Linux is pretty much owned by Large Tech Companies, just look how d

  • by demon driver ( 1046738 ) on Sunday October 08, 2023 @07:33AM (#63909655) Journal

    While I clearly am not one to whine about "compliance best practices", I don't really like the direction in which the Linux Foundation is sailing, either. As I already asked a few days earlier, how does Kroah-Hartman even know "nobody" uses those kernels? Does the Linux kernel do telemetry? Of course they're free to (not) do and (not) say what they want, but if they give an explanation like that for such a—in the literal meaning of the word—far-reaching decision, it wouldn't hurt to back it with at least a minimum of substance, either.

    As @jmccue already said elsewhere in this forum, the Linux Foundation is a large-enterprise venture practically owned by companies like Microsoft, IBM, Oracle, and if the kernel people would understand the world's economic operating system only half as well as they understand their computer operating system, they wouldn't have fallen into the delusion that especially Microsoft would really have the best interests of Linux at its greedy heart.

    I'd like to repeat some Linux Foundation figures which @jmccue already noted and add some more:

    2021: revenue forecast $177M, 3.4% allocated to 'Linux kernel support'.
    2022: revenue forecast $244M, 3.2% allocated to 'Linux kernel support'.

    Yes, in absolute figures that's more in 2022. But isn't 3.x% dedicated to Linux a bit sad for the 'Linux Foundation'?

    • Linus Torvalds wouldn't have released his OS under the GPLv2 if he wanted to get rich from it.
      • by jmccue ( 834797 )
        Yes you are correct, but he did get rich in the end. He turned out to be even better at Project Management then programming. And management is were the real $ is at :)
  • imagine taking lunduke seriously. buyer beware.
  • ... to send complaints EVERY SINGLE TIME *anything* breaks and make them PAY for this arrogance?

  • ...is one of public perception, particularly the perception of CEOs and CTOs who make judgements on technology with limited or no actual technical understanding.

    Linux being seen to reduce technical support for long-term kernels could easily come across to the non-technical folk as an admission that "Open Source can't hack it", that Linux is unmaintainable.

    This would not be the reality of the situation and I fully understand and respect the ACTUAL reasons for the cutback, but upper managers aren't paid to un

Hokey religions and ancient weapons are no substitute for a good blaster at your side. - Han Solo

Working...