Slashdot Mirror


Linux Kernel 4.2 Released

An anonymous reader writes: The Linux 4.2 kernel is now available. This kernel is one of the biggest kernel releases in recent times and introduces rewrites of some of the kernel's Intel Assembly x86 code, new ARM board support, Jitter RNG improvements, queue spinlocks, the new AMDGPU kernel driver, NCQ TRIM handling, F2FS per-file encryption, and many other changes to benefit most Linux users.

9 of 142 comments (clear)

  1. How many people to thank? by plover · · Score: 4, Insightful

    There's a long list of people who have contributed work to this, and I'd just like to say thanks to all of them.

    --
    John
    1. Re:How many people to thank? by allcoolnameswheretak · · Score: 5, Insightful

      Most civil first post on Slashdot, ever?

    2. Re:How many people to thank? by Kjella · · Score: 4, Insightful

      How many people to thank?

      According to this:

      Each Linux release includes more than 10,000 patches from more than 1,400 developers and more than 200 corporations.

      Of course a whole lot of them work on some driver that won't have any effect on you unless you own that piece of hardware, same with architecture-specific code and various other subsystems. The number of code changes that touches everyone is significantly less.

      --
      Live today, because you never know what tomorrow brings
  2. Their work is being wasted. by Anonymous Coward · · Score: 5, Insightful

    We're getting to the point where the Linux kernel itself is superb, but everything built on top of it is becoming utter shit. This is unfortunate, because the kernel alone is not very useful. The kernel's actual usefulness comes from it laying a solid foundation for the great things that could potentially be built upon it.

    Once above the kernel, things start getting pretty bleak. First we have systemd. Its ideological and architectural flaws are such that they cannot just be fixed. For example, you can't just apply some code changes and have binary logging start being useful. No, it's a broken concept, and thus any implementation of it is inherently broken as well. The same applies for pretty much everything else systemd does.

    Then we have the desktop environments. KDE isn't too bad, and there are some lightweight alternatives that a quasi-usable. But the former star of the Linux desktop environments, GNOME, has pretty much destroyed itself with its GNOME 3 effort. This is one of the most stunning failures ever seen when developing software. The user experience has been ruined in a way that many thought would not be possible. Yet it has happened.

    On top of all of that, we have software like Firefox. Like GNOME 3, its UI has been reworked in the stupidest ways possible, which has in turn destroyed its usability. Long-standing bugs and performance issues go unresolved while the UI gets worse and worse, and even ads have been injected into the browsing experience!

    So now we have a fantastic kernel, but a userland that's totally awful from its very bottom to its very top!

    This wouldn't even be a problem if we had some diversity among the major Linux distributions, but that has pretty much vanished, too. They're almost all using systemd by default. They're almost all using GNOME 3 by default. They're almost all using Firefox by default. The only ones that aren't, like Slackware and Gentoo, are rife with a different set of problems: they're goddamn impractical. The whole point of using a Linux distro is so that its maintainers do the work to integrate and compile everything, and provide a widely usable default configuration. Gentoo fucks up the compilation part to a large extent, and Slackware totally misses the boat when it comes to providing a usable system out of the box.

    The saddest part is that it wasn't always like this. While the kernel has typically been top-notch, the other software running on top of it used to be pretty good. There were numerous init systems, including sysvinit, that were better than systemd. The whole notion of "services management" wasn't even needed because such things become trivial when doing things the UNIX way. GNOME 2 was once a fantastic desktop environment. Firefox used to have amazing usability. Thankfully the kernel hasn't fallen victim to the mediocrity and destruction that has ruined so much of the software that runs on top of it. But this gets us back to the original problem: an excellent kernel is useless without an excellent userland.

    1. Re:Their work is being wasted. by chipschap · · Score: 2, Insightful

      The user experience has been ruined in a way that many thought would not be possible.

      Many thought not possible? Think: Windows 8. A failure far more stunning than Gnome 3, Unity, Firefox, or just about anything ever seen anywhere. Although it's apples and oranges, it makes systemd look stellar.

    2. Re:Their work is being wasted. by Anonymous Coward · · Score: 3, Insightful

      Don't forget the security aspect of things. Systemd is this large chunk of code that has network connectivity and is as close to kernel space as possible.

      Is it tested for security? No. Has it FIPS and Common Criteria testing? RedHat 6.x have, 7.x have not.

      Is it audited for security? We as admins and IT people have no assurance that the fact it listens and communicates on a network doesn't make it a big fat security hole, especially with edge cases.

      How good code is it? Nobody knows, as it hasn't been audited. It could be just as bad a code job as OpenSSL, but nobody knows.

      Is it better than modules? SVR4 has stood the test of time. /bin/su has been tested, vetted, and secure. Systemd has none of that, and hasn't even emerged into a long term version that is stable enough to certify for a LTS version of Linux.

      As an enterprise admin, I would say I really don't have much confidence in systemd, just because there has been no thorough testing and audit results. Couple that with the fact that the systemd of today is different from the systemd 2 weeks from now, and it is a platform too unstable for production work.

      systemd isn't bad. We just need something to show to auditors that it was programmed by people who know what they are doing, especially when it comes to real money at stake if there is an outage or a security breach.

    3. Re:Their work is being wasted. by Anonymous Coward · · Score: 1, Insightful

      Although I'm a long time Linux user, including having used a number of distros that use systemd, I've also had to use Windows 8 a lot, on many different systems. I've never once had Windows 8 fail to boot properly due to a software issue. But on numerous occasions I have had Linux systems fail to boot due to various problems involving or affecting systemd. A boot failure is the worst kind of user experience failure that an operating system can suffer from: it inherently means that the OS is unusable!

    4. Re:Their work is being wasted. by arglebargle_xiv · · Score: 1, Insightful

      The entire FreeBSD ports tree had in Q1 2015 a bit less than 7000 commits from 163 developers

      That's because they have a small, focused developer group, not a global clusterfsck where everyone gets to stick their dick in.

      (Yeah, I know, this'll get modded flamebait, but I'm trying to make a serious point, a small focused group typically does a much better job than a Monglian horde).

  3. Increasingly irrelevant by Anonymous Coward · · Score: 1, Insightful

    With the growing number of distros embracing systemd and systemd's continued march to get its tentacles deeper and deeper into the dependency tree such that it becomes a de-facto "requirement" (and once it's in your house it stays forever, and the next update includes who knows what), is Linux the kernel on its way out? I don't understand why Linus is not more concerned about this than he seems to indicate. People are ditching the system his kernel powers (and therefore his kernel) because of this. At the moment there are enough people still running older systems (e.g., Ubuntu 12.04LTS, 14.04LTS (and variants)) that it's not quite at critical mass yet, but the time will come soon enough that those of us still hanging out on those systems will have to vote one way or another with our feet. I don't want that, but it seems like it's going to happen.