Slashdot Mirror

← Back to Stories (view on slashdot.org)

Linux 4.4 Kernel To Bring Raspberry Pi Graphics Driver, Open-Channel SSD Support (phoronix.com)

Posted by timothy on from the is-it-big-and-fancy-yet? dept.

An anonymous reader writes: Linux 4.4-rc1 has been released. New features of Linux 4.4 include a Raspberry Pi kernel mode-setting driver, support for 3D acceleration by QEMU guest virtual machines, AMD Stoney APU support, Qualcomm Snapdragon 820 support, expanded eBPF virtual machine programs, new hardware peripheral support, file-system fixes, faster SHA crypto support on Intel hardware, and LightNVM / Open-Channel SSD support.

67 comments

  1. Qemu 3D Acceleration by Anonymous Coward · · Score: 4, Insightful

    This is actually kind of cool, especially if the performance isn't terrible.

    1. Re:Qemu 3D Acceleration by Anonymous Coward · · Score: 0

      I just want to know if it's going to take longer than a year to make the necessary additions to the Windows QXL drivers to take advantage of this.

      Does anybody know of a different way to get 3D Acceleration via QEMU-KVM on Windows clients without resorting to a dedicated GPU via VT-D? Maybe one of the other video cards with a different set of experimental drivers?

    2. Re: Qemu 3D Acceleration by Decker-Mage · · Score: 1

      That's my need as well. Tons of accelerators, lots of licences, can't afford the various VDI taxes lurking out there. Intentionally. Buying Server 2008, 2012, or 2012 R2 licenses to get a VDI going is absolutely bonkers. Retail Windows 7 is cheaper for small shops. And if whatever comes out the sausage factory Windows 10 Enterprise might be retail worthy if the damn licenses make sense. It will be ever thus as Microsoft's mindset is extracting the most fees ala Oracle versus helping SO/HO and SMBs. Either accept subscriptions or just go away.

      --
      "[I]t is a wise man who admits the limits of his knowledge or skill, and that pretending either causes harm." --Terry Go
    3. Re:Qemu 3D Acceleration by geantvert · · Score: 2

      Where did you get that strange idea Linux 3D performance is terrible? I won't deny that OpenSource drivers can be less efficient that close source driver especially on devices that are not properly documented and must be reversed engineered. However, if you only consider the drivers provided by the vendors, the performances are usually similar to the Windows driver.

      Of course, Windows is usually a higher priority for the vendors so there is often a small delay before a good driver becomes available on Linux for new hardware.

    4. Re:Qemu 3D Acceleration by Anonymous Coward · · Score: 2, Informative

      Considering Linux host 3D performance is terrible, I wouldn't hold my breath that QEMU guests will have reasonable performance.

      Uh, no it's not. The difference is in the noise with Nvidia drivers on Windows and Linux. I assume you're talking about that recent shit-fling from Ars Technica which also shows floating point performance drastically harmed on Linux, like an OS can even effect floating point performance (hint: they must have used different compilers). I'm sure there's software like the software Ars tested that is shit on Linux for whatever reasons, but my own benchmarks have never shown large performance differences between the Nvidia OpenGL implementations on Linux and Windows.

      I'm kind of excited about QEMU 3D acceleration, because it gives me an opportunity to add GL support to my hobby OS.

    5. Re:Qemu 3D Acceleration by Anonymous Coward · · Score: 0

      Stop using Windows?

    6. Re: Qemu 3D Acceleration by piotroxp · · Score: 1

      God thank you for this

  2. Re:2015 by Anonymous Coward · · Score: 4, Insightful

    no, new drivers get included with kernel updates. What do you think this is, Microsoft Windows?

  3. Not quite "all in" though by Anonymous Coward · · Score: 1

    I built it just a few hours ago... and /dev/video0 (or /dev/video or /dev/video[0-9]) are all missing. And mknod will create it, but it doesn't fly if you do. And dmesg shows that its not quite there. And so you have to wait till 4.4.1-rc2 or -rc3. So run an old kernel (I keep several old ones around and the last was 4.3.0). So all is well for now, we will test 4.4 kernel in another week or two. To be fair the kernel build showed that not all kernel modules were built (and that *is* quite rare).

    1. Re:Not quite "all in" though by geantvert · · Score: 1

      So webcam support is broken in 4.4-rc1. Is that on the Pi or on x86?

  4. Yeah, so? by khelms · · Score: 0

    There's a website called phoronix where you can read all the daily minutiae of changes to the Linux kernel. I don't expect to see those items pop up on Slashdot unless there is something really revolutionary or innovative going on. These changes just sound like the usual minor modifications to support newer hardware.

    1. Re:Yeah, so? by Kidbro · · Score: 3, Informative

      Meanwhile, every single major revision of the kernel has been announced on Slashdot for at least as far back as I can remember (2.2 something).
      So how you can't "expect" it to happen is a mystery indeed.

      --
      May we live long and die out
    2. Re: Yeah, so? by Anonymous Coward · · Score: 1

      If you don't want to read about such things, you're on the wrong page, mate. Perhaps Facebook or Newgrounds would suit you better.

    3. Re: Yeah, so? by Anonymous Coward · · Score: 0

      Daily Mail more like. He'd fit right in. Posting against everything because they have fuck all else to do. Somehow feeling their pathetic existence needs justifying by letting everyone know what they think about every fucking thing in the universe.

    4. Re: Yeah, so? by Anonymous Coward · · Score: 0

      2.2? So you are new here.

      And then you should probably realise that 4.4 isn't a major kernel release.

    5. Re: Yeah, so? by Anonymous Coward · · Score: 0

      Kind of like how Timothy posts everything because he has fuck all else?

    6. Re: Yeah, so? by Anonymous Coward · · Score: 0

      Considering the new features I would say it's a major release. When was the last time the Linux developers made a"major" release according to you?

    7. Re: Yeah, so? by Anonymous Coward · · Score: 0

      4.0

      It is in the build number.

    8. Re: Yeah, so? by Anonymous Coward · · Score: 0

      2.4. :P

    9. Re: Yeah, so? by StuffMaster · · Score: 1

      I think you guys are using different definitions of "major".

    10. Re:Yeah, so? by Anonymous Coward · · Score: 0

      Speaking of major revisions, serious question here. I started using Linux at kernel 2.2.14-12. Here we are some 15 years later and I have a relatively new Centos install running kernel 2.6.32-573. This story is announcing the 4.4 kernel. Have I missed something fundamental?

      Until today I've not even heard of a 3.x kernel much less 4.x. I thought 2.x was still mainstream.

    11. Re:Yeah, so? by Anonymous Coward · · Score: 0

      2.6 went up to 2.6.39, followed by 3.0. 3.x went to 3.19, followed by 4.0.

      RHEL/CentOS tends to be conservative, so while CentOS 6 runs 2.6.32, even CentOS 7 is only running kernel 3.10 - albeit heavily patched.

      Visit kernel.org to see the latest release. Read lwn.net kernel page for awesome coverage.

    12. Re:Yeah, so? by Anonymous Coward · · Score: 0

      Parent poster here, thank you for the information! LWN looks like a new bookmark for me.

  5. Systemd kernel module? by Anonymous Coward · · Score: 0

    When's systemd kernel module coming?

  6. Flash by darkain · · Score: 0

    Yeah, but how well does it run FLASH!? https://xkcd.com/619/

    1. Re:Flash by Anonymous Coward · · Score: 1

      Oh man, are you behind or what? That got sorted out at least half a decade ago. (Also, XKCD is on 1604 by now.)

    2. Re:Flash by TeknoHog · · Score: 1

      If you want smooth video playback on any platform and choose to use Flash, you're doing it wrong.

      --
      Escher was the first MC and Giger invented the HR department.
  7. systemd by Anonymous Coward · · Score: 1, Interesting

    Shut the fuck up about systemd. You all sound like old, grumpy men that can't adapt to any sort of change.

    Read up on systemd and learn how to use it. If you spent half the amount of time learning systemd as you do complaining on the internet, you might find it's actually really good. It's like SMF, but even more flexible.

    The Solaris people pissed and moaned about SMF, but now its considered a distinguishing feature of the OS, and I doubt any Solaris admin would trade SMF for SysV init.

    1. Re:systemd by Anonymous Coward · · Score: 0

      To whom exactly are you speaking to?

    2. Re:systemd by Anonymous Coward · · Score: 0

      This was addressed to the two previous posters in the comment section of this article that invoked systemd, and to the subsequent posters that will predictably do the same.

      Do you have any other questions?

    3. Re:systemd by Anonymous Coward · · Score: 0

      then why not respond to them directly?

    4. Re:systemd by Anonymous Coward · · Score: 0

      This internet thing is confusing.

    5. Re:systemd by serviscope_minor · · Score: 1

      Shut the fuck up about systemd

      Touche my man touche. You are the first person to mention systemd here after all.

      can't adapt to any sort of change

      There's a difference between "can't adapt" and "don't like the new system". "change" is not sysonymous with good, no matter what many people in the tech world desperately want to believe.

      you might find it's actually really good

      Or you might not. It has some things which are better than the old ways and some things which are worse. The people who refuse to acknowledge that systemd has any flaws at all are as bad if not worse than the people who insist the old RC scripts had no flaws at all.

      --
      SJW n. One who posts facts.
  8. Re:The Muslims by Anonymous Coward · · Score: 0, Informative

    ISIS benefits the most from anger and xenophobia against muslims in Europe. It's their goal to turn us against them, their narration involves us trying to attack them.

  9. This is all well and good by Anonymous Coward · · Score: 0, Troll

    But what contributions from women have made it into the kernel? With valued contributors like Sarah Sharp driven away, it's taking a toll on the kernel code quality.

    1. Re:This is all well and good by Anonymous Coward · · Score: 0

      Yeah, without a women there to hold the kernels hand its going to turn to drugs and prostitution!
      captcha: placenta

  10. Re:Rasperry Pi by Anonymous Coward · · Score: 0

    Except that it doesn't run linux.

  11. KMS is not something you advertise by BitZtream · · Score: 1

    Just for reference, KMS is something you shouldn't mention and hope that people forget how ridiculously behind Linux is in this area compared to ... Well everyone else that does anything other that 80x24 text.

    --
    Persistent Volume manager for Kubernetes - https://github.com/dwimsey/openshift-pvmanager
    1. Re:KMS is not something you advertise by Anonymous Coward · · Score: 0

      It's not behind. There was an active hostility towards anything graphics related ever being instituted in the kernel for the longest time. Finally people in the linux kernel community started to realise they had made a monumental mistake, and they are correcting it. Unfortunately you are wrong on your second point too, only Windows and OSX have something analagous to KMS, and that is because the whole graphics driver is in kernel. Solaris, HPUX, AIX never had kernel mode setting. Plan9 graphics drivers are in userspace (in rio) and doesn't have KMS. On virtually every platform with X11, mode setting was done in userspace by register or io space passthrough.

      The only reason things are changing in Linux and BSD space now with KMS is because it become a collosal fuckup trying to do GPU page management from userspace and synchronise it with CPU paging in the kernel, and without being able to map GPU pages into user processes other than the X11 server, it is basically impossible to do efficient 3D graphics (direct rendering).

    2. Re:KMS is not something you advertise by Anonymous Coward · · Score: 0

      "it's not behind, they're just working to catch up!"

      stupid git - you say they're not behind then say they are behind

      keep your fucking fanboy fag shit off the internet

    3. Re:KMS is not something you advertise by Anonymous Coward · · Score: 0

      Just for reference, KMS is something you shouldn't mention and hope that people forget how ridiculously behind Linux is in this area compared to ... Well everyone else that does anything other that 80x24 text.

      Funny; because for Linux the Video driver was always User-Space; and KMS allows it to be a hybrid User-Kernel Space. At the same time that KMS was being developed, Microsoft moved from a sole Kernel-Space driver to a hybrid User-Kernel Space driver too. Odd how both parties decided on the same kind of model about the same time, isn't it?

      And, FYI - Linux introduced KMS to help with better performance; while Microsoft moved to their similar design for kernel stability as video drivers (and sound card drivers) have been some of the biggest stability banshees around - even for Microsoft.

  12. Glitchy mess by Anonymous Coward · · Score: 0

    Please tell me why I would use Linux when the typical experience is like this?

    1. Re:Glitchy mess by agoodm · · Score: 1

      Because the typical experience is not like that.

    2. Re:Glitchy mess by ledow · · Score: 1

      Because that isn't the typical experience. For a start, the guy gets into driver parameters, testing kernels, and compiling from git. None of that is typical at all, hence why the Ubuntu etc. installation instructions say nothing about that. All a newbie user would know would be that some of his hardware isn't picked up in one particular distro. Maybe some older kernels work - who knows, did they test? What about the kernels in a STABLE version of a Linux distro that's aimed at first-time users? Did that work? I'm guessing you have no answer because you haven't tried.

      (And, most importantly here, if that person was you, DID YOU REPORT THE BUGS? No? Thanks a lot for smearing bad news across the Internet without even giving people a chance to fix it).

      However, it's like comparing my latest foray into IBM BladeCenter servers. On boot from a standard Windows 2012R2 install disk, they crash. That's it. BSOD and end of game. With a 10-minute BIOS boot, it doesn't matter how long you try, it still just crashes before it gets into Windows.

      You have to create the support DVD from IBM-supplied drivers yourself. They don't supply one. This involves several HOURS of downloading, lots of disk space and knowing exactly what components you have in your blade server. Down to the difference between an HS23 (1506) blade and an HS23 (8883) blade, for example.

      Then you have to burn that to a DVD (because it's too large for CD). Boot EVERY blade from that one by one, which runs off and updates 20-30 items of firmware on the blades and server and RAID card and AMM, and network card and all kinds of other internal hardware. It can take a day for a fully occupied BladeCenter. Then it lets you chain-boot into a Server 2012R2 install CD that you've got to slipstream the latest patches into.

      Then AND ONLY THEN does it allow you to boot up into plain Windows without BSOD on bootup, so you can get on with actually installing the proper Windows drivers instead of the Microsoft-supplied ones, configuring the damn thing in terms of network, storage,etc.

      Sounds like a horror story? It is. But it's got nothing to do with the Windows CD that Microsoft supply, in the same way that latest-laptop-model-with-Windows-support-only-not-yet-supported-on-Linux has nothing to do with some inherent flaw in Linux.

      Typical experience with Linux is more generally "Okay. That seems to have worked. What do I do now?" (which is pretty much Linux use in a nutshell) while sitting at an unaccelerated login screen. I know, I've deployed Linux in schools, and we do it as part of some courses, and I've shown lots of people how to use it.

      P.S. If I grab the latest laptop off the production line and slap plain Windows on it, likely I will have similar problems. How do I know? I do it all day long. Best ones are when the wireless or network just won't work, so it can't get to Windows Update to get the rest of the drivers for the machine. And sometimes even the Windows Update drivers just crash the thing or don't work at all. The newer the laptop, the bigger the problem you get.

      Would you like me to describe the problem with a brand-new B5400 Lenovo laptop I had that consisted of ONLY joining open wireless networks and not encrypted ones, where only the left half of the touchpad worked properly, the network card didn't work, and the graphics were stuck in 1998 in terms of screen resolutions? Windows is JUST as bad in similar circumstances.

    3. Re:Glitchy mess by Anonymous Coward · · Score: 0

      no os supports all hardware fagot

      'why should i do something if i can't use a certain vendor?'

      stupid shit - buy quality stuff from a vendor that has support you can't run your faggy shit os x on anything but their hrdware fag

    4. Re:Glitchy mess by rickb928 · · Score: 1

      I thought that sort of run around ended with Itanium. Wow.

      --
      deleting the extra space after periods so i can stay relevant, yeah.
  13. Re:Rasperry Pi by Anonymous Coward · · Score: 0

    What makes you say that?

  14. Re:Rasperry Pi by Anonymous Coward · · Score: 0

    fucking spammers

    take your shitposting trash to reddit or fark or some shit

    now i will not even bother to ever buy your shit product

    azulle is a shitposting scam company bunch of thieves with shit products and shit ethics and right full of cunts

    their quantum byte is made with counterfeit parts, garbage, and contains hidden malware, spyware, viruses, trojans, and is made with slave labor

    do you really think posting your shit spam ad here was a good idea?

    azulle is a shit company
    azulle is a bunch of thieves
    azulle is stealing
    azulle is a spammer
    azulle is garbage
    azulle supports slave labor

    there you go - that should get you some results in google, glad to help

  15. 2.0.36 foevah by Anonymous Coward · · Score: 0

    Anyone who doesn't agree is with Isis

  16. Re: The Muslims by Anonymous Coward · · Score: 0

    Run along now, the adults are having a conversation.

  17. Re:2015 by Anonymous Coward · · Score: 0

    No, but it's always nice to have hardware supported without needing to bother with external drivers, which always need to be compiled against the headers for the kernel that you're currently running.